Description of the Catalogue Principles of scoring hits

Experiments on the Paranormal Cognition of Drawings [25 Part II Description of the Catalogue: Principles of Scoring Hits 22 I must now give some account of the Catalogue presented in Appendix I, but before going into details there are some points of general interest and importance to be considered. Any method of assessment which depends on the comparison of one set of observations with another involves the assumption that the two sets are similar in all relevant respects except that in which we are interested, and if this assumption is false the method is invalid. In this case it is assumed that the percipients of any experi- ment, and those whose drawings are summarized in the Catalogue, are substantially random samples of substantially the same popula- tion; or at least that any differences there may be are not relevant to the issue. The 741 percipients of the Catalogue were all persons drawn from the higher educational classes of what I may reasonably term Western European culture (with which I include the United States of America)—as opposed to, say, Indian, Negro or Chinese. On a rather rough estimate, about 80% of them were students of between 18 and 23 or 24 years of age, and there was a slight preponderance of women (about 55%). The great majority were British, but there were 82 Dutch in Expts. IV and V, 12 from U. S. A. in Expt. IV, and about the same number in Expt. VI. Ideally, of course, we should prefer any experiment destined to be assessed by means of the Catalogue to consist of precisely the Same proportions of these various classes; but this will seldom if ever be realized in practice, and we must briefly consider what effects are likely to be produced by such differences as may obtain. On the most general grounds, it would be injudicious to use the Catalogue, except as a matter of interest, for assessing experiments Performed on percipients of a cultural background differing very widely from that indicated above, on mental defectives, persons of conspicuously low educational attainments, or very young children. Part from such extreme cases, however, it seems fairly clear that the effects due to differences between the percipients of an experi- ment and those of the Catalogue are likely to be small, and moreover which is important—that they will tend in general in the direction of safety, In the first place, if the population from which the sample of Percipients in any experiment is taken is liable to draw a higher Percentage of one type of object than that of the Catalogue, then it must be liable to draw lower percentages of some other type, since € sum of the percentages must be 100 in each case. hus, for a

26] Proceedings of the American Society for Psychical Resear plurality of randomly selected originals, the two effects will tend : cancel each other, on the average. In any particular case, of cours there is liable to be some residual effect, which might be appreciah misleading (in one direction or the other) if there were a stron tendency for different groups to favor different objects or types ¢ object; but scrutiny of the material fails to indicate, in fact, an considerable tendency of the kind, apart from the effects produce by the use of the originals. In the second place, it is easy to see that the effect of a differen in population is much more likely to lead to an overestimate of th expected number of hits than to an underestimate—at least on tt eminently plausible assumption, which the facts confirm, that peo Tl are more likely, under chance conditions, to draw familiar objec than unfamiliar, and provided the originals are selected only fro among objects appearing in the Catalogue. If, for example, oj Catalogue had been compiled from drawings of a population of 141 century Englishmen, we might reasonably expect that it would coi tain a higher proportion of bows and arrows than does the prese specimen, but no motor cars, radios, or telephones. If, then, a present-day experiment, we were to use BOW as an original, tf expected number of hits, calculated from such a Catalogue woul be too high, and the score correspondingly too low. Of course, we were to use MOTOR CAR Of TELEPHONE, we should get a grossl exaggerated score, since these objects, ex hypothesi, do not appe in the supposed Catalogue, but under the proviso made above th very circumstance prevents us from using them. This imaginat example illustrates one reason for adopting the principle that objee not listed in the Catalogue should not be used as originals, or least only with the greatest discretion. No doubt exceptional cases could be found, and may possib occur in practice on rare occasions; but I think the foregoing shou suffice to show that the risk of obtaining spurious positive rest due to differences between experiment and Catalogue populations one which, in all ordinary circumstances, may safely be neglect 23 A more important question, from the practical point of view, that of differences in the instructions under which the percipiem work. Evidently if the percipients of the Catalogue had been di couraged from drawing any particular type of object, while those’ some later experiment were not, the calculated expectations for su objects would be too low, and a spuriously high result would obtained, if originals of that type were used in the latter experime If, for example, the percipients of the Catalogue had been told : to draw human beings, the Catalogue would contain entries um HUMANS derived only from those who had ignored the instruction, if the prohibition were not imposed in some further experimgé while an original depicting a FISHERMAN were used, the percipié Experiments on the Paranormal Cognition of Drawings [27 would be likely to draw more representations of these patient opti- mists than a study of the Catalogue would lead us to expect, even in the absence of paranormal cognition, and there would be a risk of obtaining a misleadingly high score. Fortunately, however, the reverse is the case, for I began by imposing virtually no restrictions on my percipients, but gradually tightened up the instructions, in the hope of eliminating useless material. Ba In Expts. I, III, IV, and V, the printed instructions merely said “Certain drawings, all of which will be quite simple, will be dis- played .. . etc... . You are asked to reproduce these as well as you aa aren instructions were given verbally in the case of xpt. Il. In Expt. VI, I said “Do not draw vague scenes, elaborate interiors or geometrical diagrams,” and repeated this for Expt. VII with the minor qualification “‘unless your impressions are particularly vivid.” In this last case I further suggested that percipients should “reject images of which you clearly recognize the origin,” in the hope of getting rid of some of the too facile local associations or suggestions ; but I think the only result was somewhat to harass percipients, and I fail to see how it can have had any appreciable selective effect. Apart from this recommendation, the Instructions for Expt. VIII the results of which are discussed below, in Part III, B, were sub- stantially the same as for VI and VII. ah _ It will be seen from the Catalogue that the effect of these changes in instructions was virtually to eliminate geometrical diagrams from VI and VII, and these are excluded altogether from the body of the fe ete. The effect on ‘scenic’ items, on the other hand, seems to oa been negligible, as judged by reference to such entries as RIVER, AKE, SEA, FIELD, MOUNTAIN, etc., etc. But in any event these are not recommended for use as originals, for reasons which will be given in Section 25 below. We may accordingly feel confident that there h ann : : : ll been no changes in instructions liable to lead to spurious is eh hal highly desirable that the experimenter should a te 1s percipients the fact that the original will always depict a e ry simple concrete object (such things as ANGEL, DEVIL a E being ‘concrete’ for the present purpose), and will not their tg ‘ a scene. Otherwise they may unnecessarily reduce which can of scoring hits by drawing objects or mentioning items ris that it not possibly have been used as originals. Experience en mg ‘ Ms almost impossible to ensure unanimously intelligent tone 1s respect, no matter how carefully or emphatically Rie inain are framed; there seems always to be a few percipients Fortitude Rt alec d down such things as ‘A smell of turpentine,’ ie bourne eeling of nausea,’ or “The mentality of Professor ire a gain such aberrations are methodologically harmless and y to impose a slight additional handicap on the experiment.

28] Proceedings of the American Society for Psychical Researe 24 We may now turn to a consideration of the arrangement an contents of the Catalogue itself. I need hardly say that, in presenting it for the use of futur experimenters, I am inevitably more conscious of its defects than of its merits. One is naturally apt to forget the literal thousands 9} entries which have dropped neatly and unambiguously into thei proper place, and to remember more vividly the few score instarices in which difficulty has been encountered. Still, although I should be sorry to claim for it even an approximate perfection, I believ that it presents as accurate and exhaustive a picture of the output of the 741 percipients considered as could reasonably be expected and I should be much surprised, as well as deeply chagrined, if at omniscient auditor were to tell me that it is likely appreciably t mislead an experimenter so often as once in a hundred experimen 8 As regards the general arrangement, I have already explained, i section 10 above, my reasons for grouping together the percipient: of the first five experiments; and also the difference between a and | entries, namely, that the former refer to cases in which the object concerned was the sole constituent of the drawing, and the latte - to those in which it was accompanied by others; but the term ‘vir tually sole’ needs some amplification at this point. By it I mean thal I have ignored various minor items, not uncommonly met with which have manifestly been inserted as incidental trimmings or thi like and have no claim to be considered as an essential part of what the percipient wishes to proffer as his attempt at scoring a hit. Fo: example, many boats are adorned with a small flag, scenes oftet include conventionally indicated birds, and so forth. It seems to b absurd to suppose that such ancillary trivialities are to be regarded as serious attempts to identify the original, and they have accordingly not been listed; but inasmuch as it is difficult to draw the line it certain cases, or to set a limit to the extent to which a composit drawing may be built up round an item as nucleus which may appea’ of very secondary importance when it is finished, I have tended always towards inclusion rather than exclusion. I have throughou borne in mind the ruling principle that the one fault the Catalogue must not on any account possess is that of tending to cause thé experimenter to underestimate the expected chance performance 0 his percipients, and I have therefore consistently overcatalogued rather than undercatalogued, so to say, in cases of doubt; but I de not suppose that these amounted to so many as 1% of the whole. 25 The reader will notice that the cross-headings of the Catalogue ie., the names of the various objects, are set in three sorts of type namely SMALL ROMAN, ordinary, and italic. I use the first of thes for objects which I regard as entirely suitable for use as originals or substantially so, and of which the frequencies entered are cof respondingly suitable for use in assessment; the ordinary type 1 Experiments on the Paranormal Cognition of Drawings [29 used for subheadings of these, or for descriptive matter, etc.; the italic indicates that I regard the objects, etc., concerned, or varieties thereof, as unsuitable for use as originals, though only those of the type discussed immediately below should be regarded as definitely illegitimate. Rigi oe ult The reasons for this distinction are as follows: I have thought it proper to give all the objects drawn or mentioned, subject to the reservations made in the preceding section or below, insofar as is humanly practicable; otherwise it might be suspected that some process of wish-determined selection had been used tending to make the use of the Catalogue more ‘profitable,’ so to say, than it should be. But this policy of completeness, if unthinkingly pursued, might easily lead to absurd and deceptive results, as a simple illustration will show. Suppose a percipient draws and mentions ‘rolling clouds’: clearly the above policy requires that I enter this under cLoun, and rightly so, for there is unimpeachable evidence that the idea of ‘clouds’ was prominently present in his mind. Another percipient draws a horse under a tree, and vaguely indicates a suggestion of clouds in the sky. Are these also to be entered under cLoun, despite the fact that what the percipient wished to present was obviously a horse, or perhaps a tree, and that he probably was not thinking of a cloud in any serious degree at all? Again, a percipient will occasionally specifically mention the sea, or a field, in a manner which the policy of complete- ness forbids us to ignore; but almost every boat or ship is depicted more or less overtly on a sea (or possibly a river, etc.) of sorts, and every horse, cow or sheep in a field. In view of the fact that the percipients of these experiments were not instructed (as I now strongly recommend) to state explicitly what their drawing was intended to represent, it is not possible to draw the line between the cases in which sea, field, etc., was the primary and intended object, and those in which it was secondary or purely incidental. These objects: are accordingly unassessable, and therefore barred from use as Originals. Similarly, since the frequencies of their relevant occur- fence are not known with precision, these must not be involved in any process of assessment based on the Catalogue. I have not distinguished these definitely ‘illegitimate’ items by the use of a separate sort of type, and the experimenter must use his Own judgment and common sense in identifying them if occasion arises—which will be easy enough if the above principles be borne m mind. But a careful count, extended for safety to cover a con- Parable number of doubtful cases, gives the figures shown as ig feitimate or Doubtful’ in the Summary of the Catalogue ; and it bi A subtracting these from the totals for ROMAN and ifalic§ com- hed that the figures for ‘Net Relevant’ Items are obtained. 26 * have also relegated to the italicized class a considerable number a miscellaneous entries which I consider to be unsuitable for use nginals on various grounds other than those just discussed.

30] Proceedings of the American Society for Psychical Resea : Experiments on the Paranormal Cognition of Drawings [31 The second of these, alidade, is a good example. I doubt whethe one person in a hundred knows what an alidade is, or what one look like, or would recognize one if he saw it. It would accordingly } absurd to use it as an original, except perhaps on the completely unwarrantable assumption that, if paranormal cognition occurs q all, it must be of such a type that the percipient is able accurately t copy a drawing of which he does not recognize the content. whole technique—or, indeed, any other of empirical type—is base on the assumption that an object has as good a chance of being drawn, so to put it, by the percipients of any experiment as by those with whom they are compared, and this would be falsified if original; were used depicting. objects known or familiar only to a restricte class of percipients of whom a few happened to be included in th Catalogue. To do this would risk improperly handicapping the per cipients to no good purpose; but we could only improperly aid then by inserting in the Catalogue, or selecting for use as originals, object which the Catalogue percipients have not drawn, but which it i believed that future percipients are likely to draw; in this respec the position here is analogous to that discussed in section 22 above On the other hand, the method also assumes that the percipient of any fresh experiment are given, or give themselves, no bette chance of drawing any particular object than those of the Catalog had. But this would be falsified if the latter had been allowed a encouraged, by differing instructions or otherwise, to draw so larg a proportion of a type of object which the former are not allowec to draw as to make their average relevant and comparable outpu lower than that of percipients working under altered instruction: But this is not the case, as pointed out in section 19, and in any event could be dealt with by the adjustment therein described. As ii happens, the average number of items produced per Catalogue per cipient is somewhat higher than that to be expected in fresh expert ments, so that the straightforward use of unadjusted Catalogu frequencies gives a margin of safety in all ordinary cases; but an adjustment has to be made, owing to an abnormally high (or low) output by the percipients of any particular experiment, the total fo the Net Relevant entries (8,463) should be used, as the value of in section 19, not that for all entries. I have accordingly italicized a considerable number of unduly rare technical, obscure, recondite, vague or ambiguous objects, whic experience and common sense suggest are unsuitable for use ¢ originals. It is not necessary that anyone should agree with me in a these cases, or everyone in any of them and no doubt some of thet will appear arbitrary; but I think the reason for most of them w be clear enough on reflection. There is, however, no methodologica reason against any experimenter’s using them for special enquiries he sees fit to do so, except, of course, for the definitely illegitima’ items discussed in the preceding section. It must be remembered that although italicized items may be uns able (in my judgment at least) for use as originals, on one gro if or another, they are none the less, in most cases, perfectly legitimate objects for percipients to have drawn (or mentioned) if they hap- ened in fact to come to mind; the only exceptions are those (notably of a more or less ‘scenic’ type) which percipients could not have drawn if they had adhered to the instructions given in the later experiments (Cf. section 23 above). All other items, however ‘unsuitable’ on grounds of rarity, oddness, etc., must be retained in any estimate of the average output of percipients, etc., such as that used in section 19. It should be noted, however, that a modicum of common sense may occasionally be needed in dealing with these two types of entry. For example, I list ARROW, Full and arrow, Partial, and italicize the second of these. This is because I have necessarily counted conven- tional indications of the “barb and shaft’ type (i.e., > or approximate- ly so) as hits on ARROW, and felt it would add to the interest and completeness to show the frequencies separately. Now, it would clearly be foolish of the experimenter, if he decides to use ARROW as an original, to draw only a part of the object, and that is why I have italicized the Partial entry; but in assessing hits we must cer- tainly count these frequencies, since most of them were evidently intended to represent arrows. But cases such as this are very rare, and I think the individual experimenter will have no difficulty in identifying and dealing with them. 27 Two further points may be noted before I go on to describe the rules which governed the compilation of the Catalogue. First: Many of the subheadings are given as a matter of interest and for the possible convenience of future experimenters rather than for any methodological significance. For example, I list open and closed UmBRELLAS separately, because an experimenter who has satisfied himself that an effect of some kind occurs might well wish to enquire into such points as whether the use of an original depicting an open umbrella tended to elicit a higher proportion of drawings showing the object open, as compared with closed, than occur under chance conditions. The same applies to cups with and without Saucers, to open and closed boxes, and the like. But I think we Should naturally and rightly accept any umbrella (open or closed) as a hit on an original UMBRELLA (closed or open) in the absence of knowledge on the point, though it would be legitimate, if rash, to declare to win,’ as racing parlance has it, with either one before vcing the drawings; and we certainly must do so (or unfairly oe, the percipients) if any percipient merely mentions ‘an ™Mbrella’ without drawing it and without specifying its condition. aturally, we must then count both kinds in determining c. el, therefore, the frequencies of subheadings should be a Sy assessing results, though not, of course, if they refer to be edly different objects, such as Bow (archery) and Bow (ribbon) SHIP (sailing) and surp (steam).

32] Second: Probably the best way, though somewhat laborious, | tackling the assessment of an actual experiment is to have a Ii made—preferably by someone who does not know what original have been used—of all objects drawn or mentioned by the pe Proceedings of the American Society for Psychical Resear cipients, showing the frequency of occurrence of each and also if Catalogue frequency. It can then be decided—again preferably 5 persons not knowing the originals—on the principles discussed j II.B. below, and with such consultation, etc., as may be necessary which of these objects are to be deemed hits on the originals use This will not solve the difficulty of deciding what are to count hits in doubtful cases, but it has the merit of getting the initial clas: fication objectively carried out before these questions arise. Mor over, such lists will naturally form the nucleus of a ‘local’ catalogy and there can be little doubt that compilation of a catalogue calculated to give the experimenter an understanding of the kind’ things that people are in practice likely to draw, and an insight in the kind of lines on which their minds work in choosing them, whi it would be difficult to acquire in any other way, and is likely to” helpful both in forming decisions in borderline cases and in detectii apparent anomalies which may be worth following up; in other worc it is the best way of ensuring that practical first-hand experience the subject which no set of rules and formulae can ever fully repla 28 Apart from the exceptions noted above, from such others as mi be given below, and from a few negligibly rare cases in which pe ticular local circumstances warrant and demand special treatmet the following rules and principles have been observed throughoi the compilation of the Catalogue: 1. Every object drawn or mentioned has been entered; there i been no omission of objects just because I thought (as often enoug! that they were stupid things to draw, or because they were unsuital for use as originals. In particular, everything specifically mentio has been listed, even if not drawn or drawn only trivially. But mé tions of mental states and feelings, abstract qualities, etc., colors a materials are not listed. Indeterminate scrawls, etc., and geometri diagrams, patterns, etc., have been counted but not classified; th are of course italicized, and will be found in the Summary. 2. Nothing is listed twice, except in so far as it may have be necessary to enter two objects of the same ‘genus,’ so to say, but different ‘species’ under two subheadings. Thus if a percipient dra two trees of the same type in a single set of drawings, only 0 would be entered; but if he were to draw one palm tree and one tree, we should have one entry under TREE, Palm, and another une TREE, Conifer. 3. In general, a plurality of objects of the same kind is treal as a single object of that kind; e.g., a drawing of three cats is trea as if it were a drawing of a single cat, and gives us only one efi [33 under caT. Such pluralities are only given a separate heading in special cases, as where it might be convenient to distinguish the jural from the singular (e.g., single sheep and flock of sheep) or where the plurality is inadmissible or doubtfully admissible as a hit on an original depicting the single object (e.g., mention of a street as a hit on an isolated HOUSE). Experiments on the Paranormal Cognition of Drawings 4, In general, where common sense requires it, parts of wholes have been taken as equivalent to the wholes of which they are parts; eg., a horse’s head is entered as a hit on norsk, the corner of a table as one TABLE, the bows only of a ship as one sup, etc. But a draw- ing of a nose, or of an eye, ear, hand, foot, etc., is not entered under HUMAN BEINGS, but under NosE, EYE, etc.; similarly, drawings of windows, chimneys, doors, spires, etc., drawn in isolation give no entries under HOUSE or CHURCH but only under their own headings, Conversely, drawings. of men and women, who may be presumed to possess noses, etc., or of houses, to which windows, chimneys and doors are accessories that may be taken for granted, give no entries under these headings for the associated or component parts, but only under those for the objects as wholes. _ In short, parts of the body, or accessory parts of wholes, occurring in natural conjunction with the wholes of which they are parts, are not separately listed unless specially mentioned; and not even then in the case of parts of the body. The chief reason for this last conven- tion is that many percipients of the early experiments indulged in more or less elaborate descriptions of imaginary persons, usually such as they conceived to be associated with the experiment, and it 1s not practicable to draw the line between such effusions and genuihe impressions directed at the identification of the original. 3 Similar principles have been applied to garments, etc., and these—in Particular, hats, coats, trousers, dresses, boots, shoes— drawn or indicated as part of the normal clothing of human figures, are not separately listed unless specifically mentioned, except in two “4 = rare instances where the garment, etc., has been drawn with . emphasis as compared with the rest of the figure, etc., as to be ntamount to special mention. x have never paid any attention to the shape of the drawing ed yaa that it is necessarily by the shape that one decides ie, abe is intended to represent—unless, of course, the ee i Happens to have written this down. In one or two cases, though : foe COMET, GRINDSTONE), I have made entries even ii ccent gees whether the drawings were really intended to itresistiht € objects in question ; but in each such case the drawing Fesembled i: ooo the object immediately on inspection, and bo ae So closely that it would have been unreasonable to refuse a it as a hit if the object had been used as an original. I PD imac, Saale an indeterminate scrawl as an object because to somethin. Ive eye it ‘looked rather like’ or ‘might be meant for’ § or other. The procedure has always been to inspect the

34] Proceedings of the American Society for Psychical Researg drawing, decide what it represents in the plain common sense meag ing of the term, and enter those objects and those only. Fancifi interpretations of the ‘puzzle picture’ type have been absolutely an rigorously barred, and if I could not easily decide what was repr sented, the drawing was entered as Indeterminate, e.g., if the drawir might equally well have been a stone or a potato, a chopper or fla 7. If, on the other hand, it is the percipient who is in doubt ¢ to the nature of his impression, so that -he says “An X or a Y,” have made an entry under both headings, because it is clear that h must seriously have entertained both ideas. Occasionally it happe that a percipient draws one thing and describes it as another, eg a drawing of a gimlet is described as a corkscrew; here again, a for approximately the same reason, an entry is made under e heading. 8. Mention of “a picture of an X” or “a model of an J count as if an X had been drawn. 9. Composite drawings, as already implied, have been ‘dissecte into their component parts. Thus a drawing of a bowl of fruit, which apples, bananas and grapes are distinguishable, will gi entries (all 8 of course) under BOWL, APPLE, BANANA and GRAPE but if the percipient were to write down “A Bowl of Fruit” withe drawing anything, or if the kinds of fruit drawn were indistinguis able, we should have entries under nowt and Fruit, Indt. only. 10. In cases where there are two or more subheadings for | object, none of them being ‘Unspecified,’ and a percipient mentio an object of that general sort without specifying the particular vari he has in mind, the entry is made under the commonest heading. I example, a mention of “A lamp” would be entered under LAX Table, various. This convention can clearly have no effect on ca in which the frequencies of the subheadings are pooled, but v introduce a slight bias in favor of the commonest type of obje there be a question of comparing types, e.g., the open and clo umbrellas mentioned above; but such cases are very rare and ma’ think, safely be neglected for all ordinary purposes. 11. In the absence of positive identification by the percipi which I think should henceforward be insisted on so far as possible fairly often happens that, although there is little doubt as to | general character of the object drawn, e.g., a helmet, it is diffic to be sure just what variety is intended. In such circumstances have often used the sign ~, meaning ‘approximately,’ ‘substantial ‘as nearly as can be judged,’ or words to that effect, or the © responding adverbial forms where the context requires. The use this is not to be taken as implying that the object is indetermina but only that its precise nomenclature is in some doubt. I have occasionally used it in other connections, of which the nature she be clear from the contexts. The foregoing does not purport to be an exhaustive atcoutt how every doubtful point has been settled; but I think it covers Possible obje tentative an Experiments on the Paranormal Cognition of Drawings [35 principal conventions adopted, at least sufficiently to enable an experimenter to understand the Catalogue and use it intelligently, and to compile a substantially comparable one of his own if he so desires. I should like to emphasize again, however, that the proportion of cases requiring the application of these rules and conventions is small, so far as most of them are concerned, compared with the great majority in connection with which no question presents itself. 29 The reader will notice that I have grouped all varieties of birds, flowers and human beings together under appropriate subheadings in the three cases. I have not done this for various other categories such as animals, insects, fruits, etc., because it seems to me that, whereas in certain circumstances it might be desirable, or at least permissible, to pool all entries of birds, or of flowers, or of men or women, one could hardly wish to do so for the other categories. Percipients are, in actual fact, very likely to draw vague, unspecified flowers, birds, men and women, whereas indeterminate animals, insects, etc., are rare; and this confirms my own impression that from the relevantly psychological point of view, animals, etc., are much more sharply differentiated than are birds, flowers, etc. In this connection, I think I should say that the classification of flowers is by no means perfectly reliable. In cases where the per- cipient has named the flower drawn, there has, of course, been no difficulty ; and a few, such as tulips or daffodils, have usually been easily identified. All doubtful cases were reviewed by a comparatively expert horticulturist, who was able to identify a considerable number of them with reasonable assurance—at least in the sense of saying 9m the drawing appeared to represent; these were entered under the appropriate headings, and the remainder relegated to the “Indeter- minate or Unspecified’ class. But it is quite possible that in some oo the percipient had no more than ‘a flower’ in mind and flo e. Heppened to draw something which looked like a particular ie cases of birds and human beings, however, all particular ‘eles were either explicitly named or virtually unmistakable. 30 eer concluding this necessarily somewhat fragmentary account a Catalogue, there are two further points : Irst: Although the Catalogue has been compiled with the greatest ctivity, and attains, I believe, a high degree of accuracy, an its ea agit at any rate in detail, as more than : experimental. I should accordingly be glad t iv a, : gly be glad to receive 88estions for the improvement of future versions (if any), - O not reg: €s Pectally such as may be prompted by actual experience of its use.

36] Proceedings of the American Society for Psychical Resear Experiments on the Paranormal Cognition of Drawings Second: The Catalogue is primarily intended to enable fu experimenters to assess the results of their experiments ; but it seen worth while to point out that there is a certain amount of explorato; work which the interested reader might carry out with the aid of y more than the data presented in it. For example, we might set out { enquire whether objects related in some defined way to those use as originals—as by being commonly associated with them, or f assonance of their names, or otherwise—tend to appear relative! more often in experiments in which the related originals are use than in others. The question of what objects are ‘commonly assi ciated,’ etc., should preferably be determined by reference to sor external work on the subject, and some care would have to be exe cised, notably in avoiding pairs of objects so related as to be ve likely to appear in a single drawing (e.g., Blackboard and Ease Locomotive and Train); but the results might be very interestin; I mention this possibility, without elaboration, purely as a passit suggestion to any enthusiast who may be precluded by circumstane from undertaking first-hand experimentation. Part II b. Principles of Scoring Hits 31 I must now discuss the question of what principles should be adopted, and how much latitude, if any, we should allow ourselves, jn the matter of counting hits in doubtful cases. It is necessary to do this at what, in a sense, is a wholly dispro- portionate length: necessary, because otherwise many experimenters will almost certainly count hits where they should not, and thus tend to generate spurious results; disproportionate, because the actual percentage of doubtful cases is small in practice, as in the matter of atalogue entries. Logically, the position seems to be broadly as follows: The experi- menter draws an original having certain attributes a1, ao, ag... etc., and our business is to enquire whether the production by the percipients of drawings having certain attributes by, bo, bs, .. . etc., s independent of this fact and such as may be attributed to chance alone, or whether there is observable a non-chance relationship, such as is commonly accepted as a causal connection, between the two events; in particular, we enquire whether, if the original is drawn at time fo, and pp is the probability of a drawing having the ‘b’ attributes occurring at time ¢, the value of py is independent of ot. To do this for any particular experiment in which the original s used we must ascertain, first, the number, h, of the percipients of that experiment who make drawings having the ‘b’ attributes and, second, the number, c, of some sample of N percipients, who make lrawings having these ‘b’ attributes when the original is not used. f we cannot count the numbers A and c with precision our enquiry will be invalidated. The first point to note here is that it is not logically necessary for all, or indeed any, of the ‘b’ attributes to be identical with all or any df the ‘a’ attributes. That is to say, we are perfectly entitled, so far aS the logic of the method is concerned, to enquire whether the use >. 40 original representing a TIGER prompts a non-chance production “tawings representing lighthouses; it is only common sense— hich I Suppose means here our accumulated experience of more br less relevant phenomena generally—which suggests that it will probably ‘be more profitable to test for the production of tigers. qually, we are logically entitled—though it would hardly be more oto go to the opposite extreme and interest ourselves only a. where every attribute that could possibly be predicated of aged should also be predicable of the drawing—i.e., insist on ie ect Point-to-point correspondence (such as could never occur a actice) between original and drawing before admitting a hit. f ‘a whether we adopt either of these fantastic extremes, or some Feasonable standard between them, it is always imperative

t f 38] Proceedings of the American Society for Psychical Researe that we should be able to ascertain exactly not only the number percipients making drawings with ‘b’ attributes in the experimer under examination, but the number who have done so in the sampl with which the experiment is compared, i.e., among the percipien| oa of the Catalogue; we must be able to enumerate all drawings havin ‘b’ attributes which have been drawn by these percipients. If y cannot do this, assessment becomes impossible and drawings of th type in question may not be counted as hits. 32 The next thing to note is that, as common sense would sugges so experience confirms that even approximately facsimile reproduc tions of the original seldom if ever occur. An original of a le HAND, for example, drawn with fingers spread and palm downward is likely to elicit a number of hands exceeding expectation; but, I observed in my first paper (1. p. 130), “it is as if the percipie were told ‘draw a hand’ rather than ‘copy this hand,’ for we gi left hands and right hands, open hands and closed hands, apparent quite indiscriminately”; in other words, it seems to be the ‘idea’ ¢ content rather than the visible shape or form of the original drawin that is cognized, and then set down in whatever form happens t appeal to the percipient. I do not propose to discuss in detail here just what is to be unde stood by the word ‘idea,’ but we may take it to refer, in the case | any given person at any given time, to that group of images whi is brought into or nearer the center or focus of the field of conscio ness by any appropriate stimulus, such as mention of the name an object; in most cases, I suppose, this will more particular involve a visual image or images of some typical example of the class of object, or of one which has impressed itself with especii vividness on the mind concerned. But inasmuch as different peop! have different experiences, while it is from these experiences thi their ideas of objects are derived, it is wholly natural that A’s id of a hand should be different from B’s idea. Thus the view th drawings of hands, etc., in different positions, etc., are equally to regarded as hits on an original HAND, involves nothing at all in plausible, but rather the contrary, unless we commit ourselves advance to the theory that the percipient must in some sense ‘sé the original and copy it. This would be most improper from a logic point of view, and even if it were true that the percipient ‘saw’ t i he might still prefer to render it in a somewhat differet orm. Thus the first principle in scoring hits is that “An X is alway an X”; that is to say, a drawing of any object is always a hit ol an original of the same object, regardless of the position, etc., # which it may be shown. A cat is a cat, whether standing, sittin lying, or running, and a motor-car is a motor-car, whether facif right or left, towards or away from the observer. Experiments on the Paranormal Cognition of Drawings [39 33 The second principle, which is even more important and should be graven deep on the heart of every experimenter, is “A Y is not an X,” and in general, unless a drawing plainly and unambiguously represents the same, or substantially the same, object as the original, a hit may not be scored. ; It is, of course, through the small loopholes provided by the words ‘in general’ and ‘substantially’ that all our difficulties enter. They may, I think, be completely excluded by laying down and adhering to a categorical and inflexible rule that only the most direct and literal hits shall ever be counted at all—provided always, in view of the preceding section, that the literality is of idea or content and not of form. For my own part, I have no quarrel at all with anyone who chooses to adopt such a policy of ultra-literality, which can only retard and not prevent the emergence of significant results; for I am very much more afraid of uncritical enthusiasts discrediting the tech- nique in particular, and thereby the subject as a whole, by claiming illegitimate hits, than of purjsts who, at worst, can only waste a certain amount of valuable material by rejecting hits which might reasonably ibe accepted. Indeed, so long as any given experimenter is in doubt as to the reality of the phenomena in general, and is concerned primarily with strict tests of their occurrence, a policy of literalism can hardly be too strongly advocated; and even those who are confident that paranormal cognition does occur should be ex- tremely cautious until they have gained considerable experience, and should relegate doubtful cases to the cold-storage of the note-book rather than draw hasty conclusions from them. The question then arises, “When is an X not an X?” I think the answer must be based on the simplest and most common sense considerations, and particularly on whether the names of the two objects in question are commonly held to carry the same connota- tions. Any cat, in any position, etc., is a hit on cat but a tiger is not a cat except to zoologists, and it is mere quibbling to pretend that it is; and a polecat has such different connotations from cat that I should not count it as a hit. A horse is a horse, and so is a mention of ‘two horses,’ because the idea of horse must have been very much in or near the percipient’s mind; but a donkey is not a orse, nor is a mule, though I think a pony might be. An ironing oard is hot an iron, a saw-horse is neither a horse nor a saw; “a ane ship” is not a rowing boat, though I think we should rather ah Bay concede “a rowing eight”; a cap is neither a hat nor the ie but I think “a steel cap” would score on the latter. On Bat eal hand, an alligator certainly is a crocodile, for both the ben and connotations are indistinguishable for most ordinary i a . ut a lizard is not. Homonyms, of course, are entirely out tain! except for ad hoc research; that is to say, we should cer- . oh Jae count, in the first instance, an archery bow as a hit on wheth ot ribbon, though we might think it worth while to inquire er the use of the one tends to elicit the other.

40] Proceedings of the American Society for Psychical Researe| I think there are only two possible kinds of exception to this ri to show that most superficially doubtful cases can be easily resolve on common sense lines. Sometimes, however, cases of ‘real difficulty are bound to arise; for instance, is a flock of sheep a hit on SHEE (single)—or is a ram, or a lamb? I should say myself that a ra is, on the ground that most people do not distinguish ram fro sheep in their thought; but that lamb is not, because it carries di tinct connotations (e.g., of spring, friskiness, etc.) of its own. should also rule against a flock of sheep, at least for test purpose, partly because it seems to me connotationally different from a sing sheep, but mainly for the sake of consistency; for I should certain not regard a wood or forest as a satisfactory hit on a single TRE But such genuinely doubtful cases as this can always be dealt wit by invoking the principle of literalism to ensure safety. 34 It follows from what has been said above, notably in section 3] that hits must seldom if ever be scored on the ground of the shap or form of the drawing alone, i.e., on the ground that although th drawing represents a Y (or perhaps nothing identifiable) it look more or less like an X, as a drawing of a pond fringed with reed; might look rather like a crown. The reason is that we have no mean of telling how many drawings which might equally plausibly be hel to look like X have been drawn by the percipients of the Catalogue so that the essential basis of assessment is lacking. This rule i extremely important and must be closely adhered to, though it bear hardly on certain percipients (and these, by inspection, by no mean the least gifted) who seem correctly to cognize the form of th original, or part thereof, without being able to identify the content I do not think that this means that they in any sense ‘see’ the original but rather that images of the marks made by himself on the pape must form part of the experimenter’s contemporary ‘idea’ of th object he is drawing—i.e., must be associated with the other image constituting that idea—and that percipients of this type happen ft select these images in preference to others by reason of som idiosyncrasy of their own. Sometimes this rule operates extremely harshly: For example in Expt. VIII, HK, mentioned below, the experimenter illustrate the word INCLINE, as an original, by a sketch of a car ascending a steep hill, and a percipient drew a skier sliding down a steep slope both sketches being markedly diagrammatic. The temptation t award a hit was almost irresistible, but I could not accept it at its face value, because the Catalogue does not show the number of steef slopes drawn by its 741 percipients. I think the foregoing examples, few though they be, should se namely (a) if the shape of the object-depicted in the original is so distinctive that no drawing of substantially the same shape could humanly speaking, be plausibly held to represent anything else an¢ would certainly have been catalogued as that object if it had occurred, Experiments on the Paranormal Cognition of Drawings [41 r (b) if we have antecedent evidence to the effect that percipients fad to draw object A when object B is used as an original. As an example of the first, I recently conceded a hit on an original GLOBE (as used for geography) to a drawing which appeared unmis- takably to represent a sphere mounted on a vertical stem carried by 4 circular base, the proportions being approximately those of the usual globe. But even here, although to deny the hit would have been to carry prudence to a preposterous length, I should have felt some qualm of conscience had I not been confident that, if any similar drawing occurred among those catalogued, it would have been duly listed, probably under ‘Lope, very approximate shape,’ as was done for the two quasi-catapults, g.v. This is, however, an extreme case, and I mention it only to show how close the correspondence between drawing and original must be before the award of a hit on a basis of form, without certainty as to intended content, becomes permissible at all. As for the second, the most striking example is the outcrop of hour-glasses in Expt. VI, which I have little doubt were determined by the original Bow, drawn as a bow tie, used in that experiment, and I strongly suspect that the same is true of the milk bottles and THERMOS of Expt. VII. In the former case at least there is not only a strong resemblance between the outlines of the two objects, but ‘the excess of hour-glasses over expectation is significant. Conse- quently, if in some future experiment I were to use Bow (tie) as an original, I should feel entitled to count as hits any hour-glasses the percipients might draw (and the same, to a less extent, with THERMOs and milk bottles), but it would be incumbent on me, if I so decided to add the net frequency for hour-glasses (viz., 5) to that for Bows (viz., 4) in determining the value of c, regardless of whether hour-glasses were drawn or not ; for it would clearly be unfair to give ourselves the benefit of them if they were drawn while evading the penalty, so to say, if they were not. The same principle must be applied, of course, in all similar cases; that is to say, when- ever for any reason at all we decide that it is proper to count an X’, say, as a hit on an original X in determining h, the correspond- mg net catalogue frequency of X’ must be added to that of X in etermining c. On the other hand, if HouR-GLAss were employed as an original, should use only the frequencies under I-V and VII in determining ¢, for the occurrences in VI are at least strongly suspect of being due to Bow, and thus not attributable to chance alone; and it is clearly imprudent to use suspect data if others are available. This, however, Should not be done in other cases merely because one of the three Pairs of frequencies shows some considerable excess over the other two, for a certain number of such excesses is to be expected by Chance in a sufficiently large sample of objects; it is permissible only if the excess is significant and there is some plausible suggestion that © Occurrences have been influenced by some identifiable factor.

42] Proceedings of the American Society for Psychical Resea 35 At first sight there may appear to be some contradiction betwe the remarks of the two preceding sections and those of section | where I pointed out that we are perfectly entitled to test whett there is a non-chance connection between the use of any origif and the occurrence of any type of drawing whatsoever; but this not the case, for section 31 deals in this connection with the abstr, logic of the matter, which is not quite the same as the prac methodology. We are perfectly entitled to decide, before we see material, to count any specified sort of drawing as a hit on a original, or to enquire, after we have seen it, as a matter of intere whether any sort of drawing occurs in excess of expectation; w we are not entitled to do is to look through the drawings and pi out as hits anything that happens to strike our fancy and looks if counting it would be profitable. To allow this would be to op the door to every kind of wish-thinking, which we must above things avoid, and that is why I have insisted on literalism of con as the basic principle which should guide us in scoring hits. I should, however, be sorry if this insistence were to be taken implying that we should never exercise our discretion or permit of selves a judicious latitude in our scoring. There seems little dou that an original may in actual fact be responsible for the producti of drawings not quite identical with itself even in content, and ignore these altogether in the interests of safety at all costs wot be to waste valuable material. Moreover, it is not always possi in practice to foresee what form such aberrations may take, so fl the experimenter may well find himself confronted with a drawi which he feels it would affront reason to deny as a hit but whic has not decided in advance to accept, simply because he has 1 thought of it. This kind of thing might lead to his obtaining too I a score, which, as an open-minded investigator, he would presumal deplore only less than a too high one. The fact of the matter is, I think, that our scoring policy, particularly the degree of latitude we permit ourselves in deviat from strict literalism! should depend on what it is that we are try to do. This may range from adducing evidence which shall be pri against any rational attack of even hostile critics, to finding hi that may throw light on the psychological processes involved suggest future experiments; for the first purpose we can hardly too literal, for the second, even relatively far-fetched possibilit may be provisionally entertained. But in general, and especially | such routine purposes as studying the relative merits of differ classes of percipients, sizes of experiments, or types of original, want the best balanced and most accurate results that a conscienti0 study of the data can give us; and these will not be obtained | a too drastic rejection of apparent hits in accordance with infle applied rules. 1 The qualifying words ‘of content’ are here and hereinafter tale understood. ; [43 There is, in fact, an important practical difference between what I have termed ‘test’ and ‘informational’ scoring. For the first, that js to say for the purpose of obtaining virtually critic-proof evidence, we must keep almost blindly to the strictest rule of literalism, pecause this is our only safeguard against allegations to the effect that our scoring has been biassed by our inclinations; but for the second we should allow ourselves such reasonable latitude as com- mon sense tempered by caution may suggest, just as we do in any other field of scientific inquiry. We shall, to be sure, occasionally make mistakes, sometimes in one direction and sometimes in the other; but this is true of any work whatever into which any kind of judgment enters, and is no reason for invariably binding ourselves by the dictates of a rigid formalism. Experiments on the Paranormal Cognition of Drawings 36 There are, I think, two principal types of situation in which a certain latitude may reasonably and properly be allowed. The first is when an object fails to conform to the rule of literalism more by an accident of nomenclature, so to put it, than because of any real difference of content or idea between it and the original. Thus a chair of notably ceremonial appearance or function (e.g., a presidential chair) is scarcely distingittishable from THRONE by any reasonable criterion, and it seems barely practicable to draw a hard and fast line between tent and wigwam, though these several objects have different names and appear in the Catalogue under different headings; but I think they should be regarded as virtually Synonymous for informational purposes. Sometimes, of course, the experimenter is forced to treat two or more distinguishable objects as indistinguishable simply because they are not separately listed in the Catalogue (e.g., open and closed motor-cars, cats and kittens) ; but this, even if he may disagree with the arrangement, can do no harm methodologically since it is in the nature of an arbitrary convention applied to all percipients alike. j The other type of situation, which is commoner and more interest- img, 1s where the original represents an object very rarely drawn by Percipients, and of a character, therefore, which does not come readily to mind in the context of the experiment. In such cases it seems to me perfectly legitimate, for informational purposes, to accept ‘near misses’ or ‘nearest approaches,’ provided they are genuinely ‘near,’ and show strong natural affiliations with the original y straightforward common sense standards. Thus, no Catalogue agi actually drew or mentioned a PRAWN or a BUFFALO, ito .. ta got so far as buffalo horns ; but I have no hesitation—for for Era purposes, I repeat—in accepting lobster or crayfish, Re that rst or bulls, cows, etc., for the second. Indeed, it seems to Bitside a would be definitely foolish to do otherwise ; for, if we tes ~ low many of the images which in association make up the Of lobster, Say, must be closely similar to, if not identical with,

44] Proceedings of the American Society for Psychical Resear those that make up the idea of prawn in the minds of most peopl it is clear that the presence of the one is almost a guarantee thg the other is very close; and this, assuming it to be a legitima inference, is really much more important than the way in which th percipient actually happened to formulate his thought. Needless t say, however, the utmost discretion should be exercised in sud cases. One small practical point remains to be mentioned, namely tha if, for any reason, the original is not a simple object, but resolvabl into two or more components each of which might form an origina by itself, then these components should be treated as separa originals in all respects. For example, the ROYAL STANDARD of II consisted of a flag plus a lion, and the rLac of Expt. IV of ; flag plus a cross (the cross being the most striking feature), and ] have acted accordingly, treating each as two originals, flag and lio and flag and cross, in the scorings of section 37 below, and similarl in one or two other cases. In the interests of simplicity, however such composite originals should be avoided. To sum up: For all test purposes the policy of literalism of con tent should be resolutely adhered to, even though it means that thi experimenter will have to do rather more work in order to obtain significant result; but for informational studies a certain amount ¢ latitude is permissible, on the lines indicated, subject to the exerci of great discretion. The experimenter should, I think, invariab indicate whether his scoring has been strictly literal or not, an wherever practicable, the nature and importance of his deviation from it, if any. periments on the Paranormal Cognition of Drawings [45 Ex Part III The Technique in Action A. Experiments I - VII 37 It would be as imprudent to proffer a technique for general use without ascertaining the kind of results it gives in practice, as to put a novel type of automobile on the market without adequate tests on the road; and experimenters can hardly be expected to embark on an even mildly troublesome investigation unless they have some fairly good notion of their prospects of success. To a certain extent this preliminary testing can be retrospective; because, on the purely operational side, I have but trifling modifica- tions to suggest in the procedures described in my earlier papers, while the method of assessment may as usefully be tested on existing data as on fresh, We may accordingly begin our review of the technique in action by using Fisher Scores and the Catalogue data to assess the results of the first seven experiments; and this will have the added advantage that it will exhibit their relative degrees of success in a way which the earlier cross-scoring necessarily failed to do. The results of doing this are given in Tables Ia and Ib below. For the first of these the most rigidly literal methods possible have been followed—indeed, literalism has ‘been carried to almost fan- tastic extremes, as by not admitting buffalo horns as a hit on BUFFALO, or field guns, etc., for SHOOTING; and the scoring adheres strictly to the indications of the Catalogue, i.e., only those objects are counted as hits which are there specified as having been used as originals—for example, only hemispherical bowls are counted as hits on the nowt of VII, E. In the second, I have allowed a modest latitude, in accordance with the principles discussed above, e.g., counting cows, etc., for BUFFALO and flocks and lambs for SHEEP; but even here the utmost discretion has been exercised—for example, ave not counted hour-glasses for Bow in VI, though I think it Would probably be perfectly legitimate to do so. The figures in I a ave ‘been computed by the ultra-conservative procedure shown in the first part of the Example, but those of Ib by the multiple unit’ method normally used. In dealing with Expts. I-V, I have been peided by the considerations mentioned in section 11 above; _that is +2 Say, I have taken N as 491 in all cases, and have made correspond- Ing deductions, in determining the values of c, of all frequencies in € first two columns of the Catalogue from the gross Catalogue Tequencies, € values given under the heading of ‘Mean Scores’ are Mean Sores per percipient per original, with corresponding variances.

46] Proceedings of the American Society for Psychical Resea A That is to say, the mean scores are obtained by dividing the actual scores by the product ms in each case, where nm is the number percipients taking part in the experiment and s is the number o originals used, and the variances by dividing the actual variances by the squares of these products. A compound original split into twe or more components, in the manner discussed in the preceding section (as has been done wherever circumstances required it) counts as two or more originals. These means enable us to judge of the relative degree of succes: of the various experiments regardless of the number of percipient taking part in them or the number of originals used; other thing: being equal, the actual scores would, of course, be approximatel} proportional to the number of percipients and the number of origina Weighted means are obtained by weighting the contribution, inversely as their variances in the usual way. 4 The indications of recent work, not yet ready for publication, ar to the effect that the low scores for Expt. VII are due to the faq that the percipients of this experiment were not provided with : photograph showing the surroundings, etc., in which the original would be produced and displayed, as was done in Expts. I and III- T 1 ABIL la LITERAL SCORING Actual Scores Mean Scores Mean Vce : Expt. n s Score Vee x 103 x 106 D/o <P I 37 12 7.212 14.619 16.243 74.157 1.89 F II 20 14 = 12.895 12.325 46.054 157.206 367 OG Ill 11 10 426 5.267 3.873 435.289 19 85 IV 105 tt 4,972 62.071 10.365 46.529 1.52 13 4 5 a 746 = 45.664 69 77.018 dha Total I-V,, 250 ,. 57 . 33.251. 139.946 2.81 01 Weighted Mean I-V 13.363 17.675 2.86 VI 246 10 34.558 125.444 14.048 20.729 3.09 —.00 VIIA 245 10 3.868 98.636 1.579 16.433 39 70 VIIB ¢ STiigpood. 1 tOn ode 11.245 17.148. - 2.72 ae VIIC “ bs 5.285 97.550 2457. 16252 54 ) VIID a “ 12.253 106.124 5.001 17.680 1.19 VIIE od i 4.301 116.920 1.756 19.479 40 } Total VII “ 50 53.258 522.163 4.348 3.480 2.33. 028 TOTAL I-VII 741. 117 121.067 787.553 4.31 —10 Weighted Mean I-VII 6.841 2.550 ments on the Paranormal Cognition of Drawings [47 Experi TAB LL Eo Db, INFORMATIONAL SCORING Expt. n Ss Score Vce Mean Vce D/o “=P I 37 12 17.030 17.932 38.356 90.963 4.02 10-4 Il 20 15 24.484 14.790 81.613 164.333 8.36 10-9 Il 11 10 4.370 5.611 38.818 463.719 1.80 07 IV 105 11% 7736 61.019 15.365 45.741 227 03 V 77 10 9.520 46.445 12.364 = 78.335 1.40 16 Total L-V 250 58 73.040 145.797 Weighted Mean I-V 27.770 18.565 6.43 10-9 VI 246 10 52482 124,565 21.212 20.584 4.67 10-5 VIIA 245 10 6.538 99.481 2.670 16.633 65 51 VIIB iM “48.074 107.013 19.622 17.828 4.65 10-5 VIIC we ZEAGY A090 227 9.021 19.863 2.02 .05 VIID # y 2B26 AG2558 11.480 18.752 2.65 01 VIIE i bs 9.897 119.805 4.040 19.960 90 a Total VII 245 50 114.736 558.444 9.336 3.721 4.86 10-6 Total I-VII 741 118 239.958 Weighted Mean I-VII 828.806 8.35 10-9 13.585 2.694 8.27 10-9 38 I must here make a digression of considerable theoretical interest concerning a point, which, if it had been correctly formulated, would have constituted an important criticism of the whole technique, irrespective of the method of assessment employed, or of almost any other depending on the use of ‘free’ material such as drawings. eink this may be stated with sufficient accuracy as follows: oc method of assessment we may adopt, we are concerned act y with the question whether certain events—notably the aie 1on by the percipients of an experiment of a certain type of a more often than we should expect on certain assump- Matic 4 ably that they will tend to produce them with the same eine pmacy as the percipients of the Catalogue. But we also — .. nen case, that these events in which we are interested influences ey ently of one another except insofar as they may be cerned t, by paranormal cognition or like cause which we are con- BF concl: investigate; and insofar as this assumption is incorrect ever, ange is likely to be invalidated. Thus, if any factor, what- ore jan paranormal cognition, so operates as to cause two Where Sep cpients to draw the same object, used as an original, Ris ie (or none) would otherwise have done so, we may group of , ee counting as plural and independent a tO wit the Ss which were in fact no more than one single event, operation of the factor postulated. This type of error,

i 48] Proceedings of the American Society for Psychical Researe which may be termed the pro pluribus unum error, has been an still is all too common in attempted applications of probability theory from the time of Condorcet onwards, and we must be most careft to make sure that nothing of the kind is responsible for the result obtained in this work. Errors of this type might arise in many different ways. Fo example, it may safely be premised that the relative frequencies wit which different objects are drawn will not in general be the sam for the two sexes. If, then, an experiment be performed in whi¢ there is a somewhat higher proportion of women than in the Cate logue, and if the originals used in that experiment happen to b somewhat above the average as regards their liability to be draw by women rather than by men—which, on a random selection o originals, is an even chance—it is likely that a certain number o hits may be scored, not because the percipients concerned receiv separate and independent impressions of the object represented, bui because they are women—i.e., by virtue of a common factor operat ing upon them—and the resultant score will be somewhat too high On the other hand, if the proportion of women in the experime be lower than that for the Catalogue, or if the originals happen be relatively less ‘popular’ with women than with men, then thi effect will be in the opposite sense and the score will be somewhai too low. Precisely similar considerations apply to differences nationality, or of age, or of any other attribute that can be predicates of percipients. 39 Closely allied to the foregoing—indeed, identical from the statis tical standpoint—is the possible influence of ‘topical interest’ ordinarily understood. One might reasonably argue, for exampl that if poc happened to be used as an original at a time when a do show was in progress, percipients would tend to draw dogs mor freely than they would otherwise have done, and that this would lea to spuriously high scores. This would be another case of the pr pluribus unum error, because we should be treating as plural am independent a number of events, namely drawings of dogs, whi¢ were really no more than manifestations of one single event, name the holding of the show, operating on several percipients. More generally, it would be perfectly correct to maintain the whereas the method assumes that, in the absence of paranormal cogni tion, the probability that any object will be drawn by a percipient constant—in particular, the same for the experiment as for th Catalogue—this will never in fact be precisely true, and that th calculated score will always be somewhat too high or too lo according to whether the average probability for the originals use happens to be above or below that given by the Catalogue. Thus } is only necessary for the critic to maintain, in the case of any givél experiment, first, that the average probability ‘just happened’ to D higher than was assumed and, second, that the effect of any suc happen ments on the Paranormal Cognition of Drawings [49 Expert ing may be a good deal bigger than we suspect, in order to for the emergence of a positive result without invoking aranormal cognition at all ; and he can do the same, mutatis mutandis, for the factors discussed in the preceding section. Criticism on such lines as these could easily be made to appear extremely formidable, nor can it be refuted merely by. arguing against it. Common sense and study of the material combine, it is true, to assure us that such factors as these are likely to operate, and in practice do operate, only to a quite inappreciable extent; but to affirm this is merely to express an opinion which, logically, the critic is entitled to repudiate. We may think it ridiculous to suppose that the percipients of the first experiment were affected by a passing wave of cow-consciousness, or that those of the fourth drew crosses in a sudden fit of religiosity, but we have no remedy if the critic chooses to aver that he finds such hypotheses much more plausible than that of paranormal cognition; and in general we cannot base our claims on a belief, however reasonably held, that the effect of any such factor can only be “a very little one.’ We can, to be sure, meet suggestions about differences of sex and nationality, etc., by showing (as happens to be true) that women alone or British subjects alone give a significant result, and that therefore the significance found cannot be due to varying proportions of these; but then the critic can shift his ground to varying propor- tions of blue-eyed, French-speaking, oyster-loving, etc., percipients, and can continue the game indefinitely. All particular criticisms of this general type can, however, be met simultaneously by pointing out that, of whatsoever nature and how- ever numerous may be the factors invoked, their combined effect is as likely to be in one direction as in the other in any particular case, provided the originals are randomly selected; that is to say, it is an even chance whether they will, on balance, operate in favor of the percipients or against them. Thus, although the critic can logically, if not very plausibly, ascribe the positive outcome of a Single experiment to the chance operation of such factors, he cannot do so for a succession of experiments, or for a significant majority of a number of them, without affirming that chance operates, in Sreater or less degree, consistently in favor of the percipients; and this is a contradiction in terms. But the results given in Table I a show that, even under the Most rigorously literal scoring, every one of the eleven sets of ten originals each dealt with therein shows a positive result. The proba- bility of such a sequence of eleven results of the same sign occurring y chance, is, of course, 1 in 1,024. € may accordingly dismiss with assurance, so far as this work ncerned, counter-hypotheses based on suppositions to the effect at the results obtained have been due to fortunate coincidences of account is co th °riginals with favorable fluctuations in their ‘popularity’ due to Variations e Part } either of topical interest or of the constitution of the ‘cipating percipients. This does not imply, however, that no

50] Proceedings of the American Society for Psychical Reseay particular experiment can ever be influenced, in either a positive negative sense, by topical or like factors; and, in general, expe menters should remember that, other things being equal, a sm quantity of material, or short series of experiments, can never yie evidence so reliable or so cogent as a large quantity or a long series. On the other hand, just as was the case with sex, etc., prefe ences, inspection suggests that topical influences are so small as | be negligible. B. Experiment VITI 40 Under the general heading of ‘Experiment VIII’ I subsume # work of 25 different experimenters, conducting independent expe: ments, between November 1940 and February 1942 inclusive. Ty of these, ‘PG’ and ‘RH,’ did two experiments each, while sever of the other experiments consisted of two or more parts. For co venience of tabulation, etc., I also include the Wing experime already discussed, which was of the same independently condu type though performed at a much earlier date. Although the work was both protracted and laborious, my accou of it must be exceedingly abbreviated, but this is no disadvanta since none but trivial alterations were made in the procedure pret ously adopted. These will, however, be duly noted below. The work was initiated in pursuance of thé policy of attempt to develop a repeatable technique for general use, with the obj of seeing how experiments with drawings fared in hands other th my own, and particularly of ascertaining whether the method assessment by cross-scoring was usefully applicable to individt small-scale experimentation. With these objects in view, I proached, in the summer of 1940, a considerable number of peop some in the British Isles and others:'in the United States of Ameri who I thought might be willing to collaborate, and invited them try experiments on these lines. I must here express my deep gratitu to the following for the help they have given me, and for the tin and trouble they devoted to the work:! ; In U. K., Dr. J. M. Thorburn (Cardiff); Miss Marjorie Hai mond (Birmingham); Dr. P. E. Vernon (Glasgow); Mrs. Oliv Gatty, Dr. C. M. Fleming (London); Miss Estelle Canziani, M G. le M. Mander, M. P.; Mr. G. H. Spinney; Dr. M. I. Balfot Miss Theodora Bosanquet; Mrs. Frank Heywood; Mrs. W. Knea 1 1 Names are given in approximately the chronological order of the expe ments. In the case of University teachers, etc., the name of the Universi follows in parentheses. : { Experiments on the Paranormal Cognition of Drawings [51 (Oxford) ; Dr. Mary Collins (Edinburgh) ; Mr. J. Fraser Nicol; Mr. Kenneth Richmond; Viscountess Traprain; Dr. Ll. Wynn- Jones (Leeds); Mrs. M. E. Bosanquet. m U.S. A, Dr. C. M. Harsh (Nebraska) ; Dr. D.. M. Allan (Hampden-Sydney) ; Mr. E. P. Gibson; Dr. H. Kettering (Colo- rado); Dr. George A. Zirkle (Hanover, Ind.); Dr. John Bentley (American Univ., Washington D. C.); Dr. Gardner Murphy (Col- lege of the City of New York). 41 Experimenters were supplied with books of foolscap size, on the es of which they were to draw the originals, very similar to those used by the percipients of earlier experiments, and bearing instruc- tions as to the conduct of the experiment printed on their covers. Three alternative types of experimental procedure were suggested, namely (A) a ten-evening experiment precisely similar to my own I, III, IV and V, (B) a group experiment, with suitable invigilation or equivalent, similar to my number I]—i.e., with all percipients present together and the work completed in about an hour, or (C) an intermediate type in which the percipients make their drawings in their own homes while the experimenter works to a prearranged time-table, producing originals at, say, five-minute intervals. The type of experiment adopted is shown in Table II under the head- ing T. Experimenters were urged to do one or more repetitions of the experiment wherever possible, or at least to prepare, immediately after the experiment proper, another series of ten ‘dummy’ originals for comparative purposes. Actually, six experiments (Nos. 2, 9, 22, 23, 24 and 26) had two complete parts each; three (Nos. 8, 17 and 25) had three parts, one of them (No. 8) having dummies for the first part also; one (No. 27) used 30 originals in a slightly different Procedure, and this may be regarded as equivalent to a three-part €xperiment ; one of the two-part experiments (No. 26) had dummies to both parts; and of the remaining 17, ten had dummies in addition to the originals proper, while seven had none. The object of these ummy’ originals (somewhat unwisely referred to as ‘controls’ in my instructions, though they could not, of course, ‘control’ any- thing) was partly to add to the interest of the work from the per- “iplents’ point of view, but mainly to provide ‘straight’ precognitive material for examination in cases where no second part was attempted. kee experimenters were directed to use the ‘dictionary’ End of selecting originals; that is to say, to insert ten numbered shuffled cards randomly into a dictionary, to open this at the a So indicated in the order given by the numbers, and to illustrate 1 ioe the first readily drawable word found. In experiments num}, » 24, 26 and 27, however, I supplied the experimenters with a €r of folded slips of paper on which I had myself written words

52] Proceedings of the American Society for Psychical Researe taken at random from the Catalogue, from which they in turn dre ten at random to determine what the originals should be. I sI a have more to say about selection of originals in section 44 below, bu it may be noted here that the ‘dictionary’ method leaves much to b desired. This is not because experimenters show any tendency t evade difficult and choose easy (and possibly topical) objects, by rather because, in an excess of conscientiousness, they try to illustrat objects which are too ambiguous, recondite or difficult of recognitio: to be suitable for use as originals at all. a ‘In the interests of paper economy, I issued percipients with doub! foolscap sheets for their drawings, instead of ten-page books. Th first sheet carried instructions, while all ten drawings were done o the second, which was divided into ten rectangles measuring abou 3%” x 2”. This seems to have had no ill effects, but rather to hat been beneficial in discouraging the production of vague ‘scenes’ af the like. The instructions to percipients were substantially the same those mentioned above, with one fairly important alteration. Insteal of telling them to ‘reproduce’ the original, or to draw the object the thought it represented, I said, “You are asked to write down” (m italics, not in instructions) “what you think each of these origina represents. Add a small sketch of your own if you can.” The obje of this alteration was mainly to avoid ambiguities and to lighten th task of the scorer by reducing the number of cases in which th percipient’s intention might be doubtful, but partly also to encourag those who could not draw, and to promote concreteness; I als wished in some degree to direct.attention away from mere shape 0 form and towards the content of the original. So far as I can judg this alteration has been beneficial, at least as regards the first poimi and I have noticed no concomitant disadvantages. It may be note that the great majority of percipients—fully 90%, at a guess—did i fact attempt drawings of some kind, even when they manifestl could not draw. : The importance of excluding all possibility of normal cogniti¢ was, of course, suitably impressed on experimenters, who were aske to sign a declaration to the effect that this had been done. Each pet cipient was also asked to sign a similar statement and one affirmin that he “was not wittingly influenced by what anyone else put down. In a few cases, where these signatures had been inadvertently omi ted, I took appropriate steps to satisfy myself that all was in orde in a few others, where I could not do this, the percipients concerneé were omitted from calculation. 42 At the time when I initiated this work I had not realized possibilities of assessment afforded by the use of Fisher Scores, afl I thought that I, and future experimenters, would have to rely 0) cross-scoring to obtain our results. I hoped, and more or less & pected, to find that if an experiment were conducted in two or mo} Experiments on the Paranormal Cognition of Drawings [53 arated by a suitable interval or intervals of, say, three a month, and all drawings were then scored against all iginals, we should obtain a positive result (not necessarily signifi- ad often enough—say, seven or eight times out of ten—to make +" procedure a worthwhile method of research. In this I was wrong, for 1 had clearly underestimated seriously the extent to which ‘dis- Jacement’ or ‘temporal dislocation of response’ may take place tween the parts of an experiment (and also, it would appear, petween different experiments ).’ It seems fairly clear to me now that astronomical time per se 1s not the only factor determining this effect, so that it is not practicable to specify any reasonably short interval between the parts of an experiment such that we may be confident that pre- and retro-cognitive effects will not appreciably erate, so to say, across it. é : J accordingly arranged for the experiments to be scored in pairs by external scorers who did not know which originals had been aimed at by which percipients. I am very much indebted indeed to Mr. G. W. Fisk, Mr. Fraser Nicol, Mr. Kenneth Richmond, Major K. H. Tuson and Bishop F. J. Western for their kindness in under- taking this most tiresome work, which is ‘not rendered the less valuable by the fact that it yielded virtually null results. : This nullity of outcome, however, relieves me of the necessity for describing in detail the methods I adopted for ensuring an arbitrary pairing of the experiments to be scored against each other—a task much complicated by the erratic and belated arrival of some con- tributions—or the devices, some of them elaborate, which I used in many cases to ensure that not even the remoter sources of error should have a chance to vitiate the results. It will be understood that, in each case, the suitably randomized drawings of all percipients of both experiments were scored against all the randomized originals of both, and the data computed in a 2 x 2 table in the usual way: similar treatment was applied to different parts of single experi- ments, in cases where more than one part had been done. _ Twelve pairs of experiments were cross-scored in this way, yield- ing eight positive and four negative results, while of the nine multi- part experiments among them six were positive and three negative. All that can be said of results like these is that they are in the right direction, though falling far short of significance ; and it became clear that either cross-scoring was too insensitive a method to be of much use for general purposes, or else the degree of interaction etween experiments, or parts of the same experiment, was likely to be so great as seriously to obscure the effects looked for. rts, SeP weeks or 43 : Meanwhile, however, the method of assessment by Fisher Scores and Catalogue data was developed, and all the material of Expt. Vill Aalihid been scored in this way. The standard of scoring adopted 1 Cf. Note to section 11 above.

54] Proceedings of the American Society for Psychical Resear Experiments on the Paranormal Cognition of Drawings [55 approximates very closely to that use for Table I a.—that is to sg almost perfectly rigid literalism—although I found it necessary exercise an occasional modicum of discretion in respect of some the less straightforward originals; on the other hand, in the in e ests of a conscientious conservatism (if this be not a contradi ic in terms) I have very frequently rejected hits which by all comm sense standards would unquestionably be accepted. : The results are given in Table II, which is to be interpreted follows: The first column gives the number of the experiment — approximately chronological order, a central date being taken j cases of multi-part experiments; the second shows the identifyi letters of the experimenter (usually initials) followed by a numb denoting the part of the experiment referred to; the third and four give the number of percipients in that part, and the type of expe mental procedure used. As regards the three main vertical divisio the figures under ‘Contemporary’ give the scores and variances f hits made by the percipients on the originals at which they we: aiming, regardless (as usual) of whether the ordinal position of # drawing scoring the hit is the same as that of the original on wh the hit is scored; those under ‘Precognitive’ refer to hits made | originals used in later parts of the same experiment, or on ¢ ‘dummies’ referred to above; those under ‘Retrocognitive’ refer hits scored on originals (including ‘dummies’) of earlier parts the experiment, but these figures have only been computed in few cases where I knew positively that the nature of the originals ¢ the first part, etc., had not been di We may say that experiments of this sara: gi type ied itive results about three times out o five, with a notice * h not significant tendency for retrocognitive data to give higher aS recognitive data lower proportions than this. It should be il ered, however, that these experiments have been to some Bent handicapped by the use of unsuitable originals, due to the tionary method of selection; and also that in some cases, where rarely drawn objects have been used and the number of percipients has been small, the total expected number of hits has been less than unity, which means that there is less than an even chance of obtain- ing a positive result in the absence of paranormal cognition. These ‘positive-negative’ data are accordingly not very informative. The unweighted mean score per ae for Contemporary data only, is 76.607/496 = 115445, sincg 496 sets of percipients’ drawings were involved, and the correspondii “mean variance (which is not, of course, the same as the variance of this niean) is 231.825/496 = 389. rnes. figures are important, for they enable us to form some idea of the amount of work the average experimenter will have to do in order to obtain a significant result by his own efforts alone—using the Catalogue, of course, for assessment. It is easy to see that a collec- tion of 100 sets of ten drawings each, of the same average merit as those tabulated in Table II, may be expected to yield a score of about 15.445 with variance about 46.739; this would give D/o = 2.26 with P less than .03, which is a satisfactory level of significance by ordinary standards. Or, if we adopt the conventional level of P = .05 and D/c = 2.0, and solve the appropriate equation, we find 78.4 as the number of sets which may be expected to yield on the average a result just attaining this level; while for P= .01 the requisite Humber of sets will be about 130. There is no guarantee, of course, that every experimenter who collects about 800 drawings, e.g., by doing four ten-original experi- ments with 20 percipients in each, will necessarily achieve a significant Fesult, any more than there is that no one who does less will succeed mM doing so. Theoretical considerations apart, a glance at the Table Would serve to disprove any such notion, for some experiments with relatively large numbers of percipients yield but small positive results, or even negative (e.g., Nos. 2 and 10), while others with small “Tumbers are very successful. But I think we may fairly say that no one who has not collected the better part of a thousand drawings, 8. five ten original experiments of 20 percipients each, is entitled S€Tiously to co on dummies are given before those on the originals of later part: thus, for Expt. 8, PV, the first line shows, under Precognitive, scores made by the percipients of the first part on the dummies the first part, and, under Contemporary, the scores made by the on the original proper of this part; the scores made by these same percipients on the originals of second and third parts; the fourth line gives the precognitive scot by percipients of the second part on the originals of the third (the were no dummies for the second or third parts) and the ‘conter porary’ score made by these percipients on the originals of their ow (second) part; finally, the fifth line gives the ‘contemporary’ sco} by the percipients of the third part on the originals of the third par A similar arrangement is adopted in other cases. The Table contains a total of 80 entries altogether, of which 5 are positive and 29 negative; these are distributed as follows: mplain if his results are not significant. tis 1S a point worth emphasizing, for I have some Teason to 4 pd st iain of the experimenters who have so a con- ie ota . <G to this work have been disa ointed and discouraged at not Precomiitive 00020) 0. 18 11 29 ews been able to detect more in signs of success than were Contemporary ove. 25 17 42 Ditmm. ' t© inspection; but, apart from the fact that it is strictly Retrocognitive secs... 8 1 9 nee to assess the results of an experiment without the aid of 51 29 80 atalogue, or equivalent body of knowledge, it seems reasonable

A GB Gt Unwan f 4 7 awe a Cte fan 56] Proceedings of the American Society for Psychical Reseap | TABBDEST SCORES AND VARIANCES FOR EXPERIMENT VIII. 2 , PRECOGNITIVE CONTEMPORARY RETROCOGN: ” No. Expr! au Score Vece Score Vce Score ord Nae ne Og fresco | eases 8.007.) SR86) Se. 3 he aber a, immer. Seana oe =a et. — 309 A! Oa 2. fF MER G37" OB 9.449 19.408 PASS TOUR, is . SU Ae ear Migr, semen abe IN, 3.988 13.554 496 13 3. go iad eo BA xe 1.289 11.135 S869. D1L205) sas 4 Me TSB — 152). 6.380) 5210 Bag] 5° Gree as) oC 142 6.852 461 5.119 aT 2 GR ots A 2.003 2.262 063 1.833 Fite: KSA 28 ot 5.678 5.654 9.278 10.818 Bon PVs Gl UB 1.184 2.520 —1.528 4.838 sea | bg 1.266 4.422 NORAD agate me: # LAT eS O00 Ts) Suita Bee Pte Las, : : 12 PB 4 DAR ES Aaa ier! 080s SBR ks MEA 2) |i aka) DAD) ZOUe 2 S00 ou by Bi oats SME au St. orate 0 +t GER eR Da nS Beier tek 10 CH ah baste Bi B67 = 82900 © 1oeie a5.a7un ae Bh = ah Cs AE Ais yar) pee eon eh Ber OT ane Pe MES) EN ian) pata ade 0) eR LEO Dons, xt Gla mee Bt ta ane ma gee SONAR. GMdnee 1 eeorecre a! t COnae Ne Demet ete Here teagan ls te MBL ee NALS Meh 2802-1 oese oS 16 EBT SB, 2D seer Aaa. oe atel ar, 1 RAS 24 OM. Peay (i ye Siar 5.737 5.443. — 633 4.803 ae ob Ak De BOOM), bi cheese etnies 4 oes san aa 1.244 4.368 5.009 2.818 “am Bane esi dUAN AB Sapttecesh mnteeae 3.935 439 @ aes PE re us eee aren 5 (mee Tee 575 5 Te MRE 12 oe ae 2.069 4.474 SAY A i pein ENE ge NEEL abba SOK, UP, ccgels Pe Gaaee DOGS 5 canasy, Bea EON ih Ce yeaa. eae rc al ae pi. ICR Be OC 3.083 1.236 BO 22°) GMA 5 B 1.063 = 1.209 AOE eile PANES Ui) ee By, Ais cyano oh eee 1.205. 3.837 En ff. see ae 178 1,098 LOSSr 21 es See amen te aa eae ee 1.098 697 1 Be NE LT 2 eG 3.011 6.082 Roser iyOt | MG a Giedeinn eee Bo N ey Ade cay8) eRe SON cn ISG ies i 669 897 Te a VA: Sas 630 896 tle) A IR © Perera ran AE 26) EBo1- 40, -B > + 042). 6246 peat i — 540 4.236 ee * —1.603 4.897 Z ‘ 10 , BO = aI7 4893 2). Reve 7 “ 2 “ fe Bop {6 eee aa : J TOTALS: 30.848 158.647 76.607 231.825 9.173 444 D/o 2.45; D/o 5.03; D/e 1.37 Pv 02; Pee 0-6 Pol GRAND TOTAL: Score 116.628; Vee 434.694; D/o 5.59; P <a Experiments on the Paranormal Cognition of Drawings [57 i t, if paranormal cognition had been so vigorous and 4 a spree eee mice as rl make itself clearly visible in a Fe ioe trials of this kind, it would have been long ago recognized a n established fact of science. As things are, it seems clear that * - wishing to satisfy himself de novo of the reality of the Ee enomenon must be prepared to undertake experiments on approxi- ely the considerable, though by no means prohibitive, scale nt indicated ; or, alternatively, he must ally himself with others sf tha their combined contributions make up an equal bulk of material. But it is also clear, which is more important from the present int of view, that, if he does this and conducts his experiments with reasonable care and intelligence, assessing his data conscien- tiously with the aid of the Catalogue, the technique is competent to provide him with the significant evidence f requires. Part IV Practical Points: Conclusion 44 Ihave an uneasy feeling that one of the effects of this paper, which is necessarily loaded with a mass of somewhat tiresome detail, may be to give the reader the impression that experimentation on these lines is inordinately difficult and complicated. This would be alto- gether erroneous, and I think the best way to counteract or guard against it will be to run over briefly the various points which naturally arise in connection with any experiment of this type, so that the prospective experimenter may realize how very little there is to be afraid of. . l. Choice of Percipients: Some percipients are, as we might expect, better than others, though I do not yet know how much tter, or how consistently. But beyond a rather faint indication to the effect that women of, say, 50 years of age and upwards are Somewhat superior to other classes, I can offer no criterion for Selecting good percipients in advance. Care should be taken to avoid Persons who are too lazy to read simple instructions, too stupid to understand them, or too irresponsible to carry them out properly. ver-earnest enthusiasts likely to force their own imaginings upon © experiment are unlikely to yield good results. Type of Procedure: There is some evidence, chiefly from Expt. I, to the effect that ‘group’ or ‘class’ experiments, in which all per- “ipients are present together, are more likely to give good results than other types; but the evidence is slender, and this type of Periment involves either invigilation by someone other than the

58] Proceedings of the American Society for Psychical Resear, J experimenter or the use of percipients who can be relied upon noj to talk to each other, or mutter to themselves, or look at each others work. All things considered, probably the ‘C’ type, in which the experi menter produces his originals at set times while the percipients make their attempts in their own homes, is likely to prove the most convenient in most cases. Whatever procedure be adopted, steps must be taken absolutely to preclude all possibility of percipients’ gaining knowledge of the originals by sensory means or by rational inference; but this is easily ) ensured, notably by always having experimenter and percipients in different rooms during the relevant periods. The need for preventing " percipients being influenced, wittingly or otherwise, by each other work has been stressed ; the signing of a declaration by the percipien that he has not wittingly been so influenced might well be made standard practice. As regards size of experiment, I am inclined to suspect that the optimum number of percipients is in the neighborhood of 20, and that performance may tend to fall off if this number be greatly exceeded. There is, of course, no a priori reason why we should no work with any number of percipients from one upwards, provided a sufficiency of drawings is collected, but I am sure that it is preferable to ensure this by doing several experiments rather than by using a great many percipients in a single experiment, while wit only one or two percipients it is largely a matter of luck whethe they happen to be good or not, and day-to-day fluctuations of ability are likely to be serious. So far as I can judge, the ideal would be a set of five experiments using fifty different originals in all and 20 different percipients in each. 3. Choice of Originals: A good original is the kind of thing one finds in a child’s painting book—a simple outline drawing of single, simple, unambiguous and reasonably familiar concrete object The object represented should be such that no ordinary person could well mistake it for anything else if encountered in the flesh, so to say; and although I do not myself believe that the marks made on the paper by the experimenter have much to do with what happens, except insofar as they force him to think about the object depicted and so to associate it with the experiment in his mind, the drawing should preferably conform to the same standard—i.e., should not be liable to misinterpretation. Although I may well have made errors of judgment here and there, I believe that most of the objects of which the names are set in Small Roman type in the Catalogue are suitable in this sense as originals, and that most of those set in italic type are not. I accord- ingly strongly suggest that, in the early stages of research at any rate, originals be selected exclusively from the RoMAN type items. It is necessary that originals should be randomly selected in the sense (a) that there must be no risk of their being determined by events of topical interest, and (b) that there must be no scope for periments on the Paranormal Cognition of Drawings [59 Ex rational inference by the percipients as ‘i pinay one “ hg bias -. more likely to be chosen than another—as there might be, f c le, if the experimenter were to take ten objects beginning with er the first experiment, ten beginning with B for the second, with Bor the third, etc. For casual experiments, any arbitrary system of selection will usually be good enough—such as inserting ten slips of paper randomly in the Catalogue and taking, say, the ninth item from the top of each right-hand page (or bottom of each left-hand e, etc.) ; but for serious and continued work I should recommend that the experimenter write each ROMAN type name on a separate small card, keep all these shuffled in a bag, and draw out ten at random and without looking at the others when an experiment is to be done. The labor inyolved is not serious and will save much time and trouble in the a run, as will also the entering of the appropriate values of c and N on the card so as to save reference to the Catalogue. In doing this I think it would be better not to make out a separate card for every subheading, even if the subheadings are not italicized. That is to say, I should use a single card for Biron, CHAIR, FLOWER, HUMAN BEING, TABLE, TREE, etc., and select the particular variety to be illustrated, if one of these is drawn from the bag, by any con- venient method. Otherwise originals are likely to include an unduly high proportion of such objects as these. a ; So far as I can see, it is nearly always legitimate to reject an original, if there is any good reason for doing so, though naturally this could be carried too far; and I think this should be done if two or more originals randomly selected happen to resemble each other unduly closely—e.g., I would reject TIGER if cat had already been selected, and so forth. Naturally, percipients should not be allowed to study the Catalogue, Or examine the cards; otherwise they might tend to concentrate on Roman type items and avoid italicized, and this would upset the whole arithmetical basis of the method. ; 4. Instructions to Percipients: Apart from inculcating, so far as Possible, an attitude of quiet confidence and receptivity, the great thing here—and experience shows that it is far from easy—is to Prevent percipients from putting down what they know must be wrong. I myself have yet to find words clear or strong enough to Induce them all to abstain from drawing or describing scenes, senti- ments, historical incidents, moral qualities, etc., etc.; but it should ~ MMpressed upon them that they are not to indulge their propensi- Hes in these directions, or to draw geometrical diagrams or patterns, and that the original will represent a single, simple, concrete, object. Cutine instructions concerning the giving of age, sex, date, etc., ay be left to the individual experimenter; but he should warn Ose who take part in more than one self-contained experiment that © Originals used, being randomly selected, may be the same in some -<S as those used in former experiments—and that the cards used any experiment should be returned to the bag, and well shuffled

60] Proceedings of the American Society for Psychical Researe into it, after each experiment is concluded; or, more generally, tha an original should not be rejected because it has been used in q earlier experiment, though the same original should not be use twice in the different parts (if there be such) of a single experimen} I think it is definitely valuable to make percipients write dow explicitly what their drawings are intended to represent. 5. Counting Hits: This is the only part of the technique in whie real difficulties are likely to be encountered; but even here they a unlikely to be so formidable in practice as my discussion may hay made them appear. In nine cases out of ten, I suppose, and probab a good deal more often than this, there will be no doubt at all to whether a hit should be counted or not, and in the few remainin cases common sense and caution should provide an answer in th light of what has been said above. To summarize the whole matter: Logically, we are entitled test for the non-chance occurrence of drawings having any sort relationship whatsoever to the original; but common sense sugges that some relation of similarity is likely to prove the most profitabl and my experience strongly indicates that similarity of idea_rat than of spatial form is of predominant importance. But whateve relationship we decide on, two things are imperative; first, t whatever objects represented in percipients’ drawings are counte as hits in the determination of h must also be counted from Catalogue in the determination of c; second, that we must not ma up our minds that we are going to count some particular sort @ object because we think it would lead to a higher score to do so, af \ this means, in general, that we must not decide on what to cou after we have seen the data. It is perfectly legitimate to say befor we see the data “We will count W or X or Y or Z as a hit on X” e.g., donkey, horse, mule or zebra as a hit on HoRsSE—but we mut not decide to count donkeys after seeing the data, just because | large number of donkeys has been drawn. In passing, I may say that, so far as my experience goes, ‘trang formations’ of this kind seldom occur; such transformations as d occur seem rather to be due to misinterpretation of a cognized shap (or rather, in my view, ‘the idea of the shape’), as in hour-glassé for Bow, milk-bottles for THERMos, amphorae for yJuG, bunch ¢ balloons for GrapEs, etc. If this be correct, it follows that to exte our range of acceptance in the kind of way just indicated will seldo: if ever be profitable, but usually the reverse, and I according strongly advise against it.} The only completely safe course is to adhere closely to the princtp of literalism (of content, not form) which has been already dis cussed, even though it will often involve discarding drawings whie common sense vehemently assures us have very strong claims 1 It should be remembered, moreover, that, if we decide in advance to cout W, X, Y or Z, etc., as hits on X, we must count the frequencies of W, Y, Z etc., from the Catalogue in determining c even if none of these appears amo: if the drawings. : ments on the Paranormal Cognition of Drawings [61 Experi ‘ ; ce, and I would go so far as to say that this course mus a" 4 for test purposes. In allowing ourselves a reasonable eS when we are interested rather in information than in test = aoe sense will usually be a safe enough guide, though itt dl uestions of real difficulty may occasionally arise. For ac pon a drawing of a Lens System be accepted as a hit on. cope? If so, would we not also have accepted it as a hit Be ccxc SCOPE. or CAMERA; and should we not, therefore, count fhe a De cies of these objects in finding the value of cP I think there ee little doubt that the answers to all three questions should be ine affirmative; but it is the kind of point which might easily ntion of the experimenter. Berricuities “of this kind ie however, extremely rare, and to say that Bey may occur is to say no more thi/n that the ge Sa Tees with almost every other scientific method the as ed ro entirely automatic, but of occasionally presenting to t ott Ww xb E accuracy cases in which the experimenters judgment mu exercised. 45 onclusion; There is no better way of spoiling a useful contri- Rison to knowledge or to method than by claiming too much for bb I am confident that the methods here described iui: ba: the Catalogue, a good and valuable technique, in the "ee t i - they are applied in the manner indicated, they will enable 5 pl petent experimenter to verify for himself the occurrence if p . normal cognition, without excessive labor, and, if he so wis iS , continue his research into the study of the kind of process that 1 is and of how it operates. But I should be sorry to suggest poms that my exposition has been perfect or that the technique, whether, in outline or in detail, is final and cannot be improved upon. ‘ On the contrary, I have little doubt that the reader will detec many faults, of omission and commission alike, in my oe which I shall be very glad to remedy if opportunity 1 aC a have every hope that we shall soon find new methods whic ie make the present look as cumbrous and as antiquated as every early instrument looks at a later stage. ok The use of drawings has at least two advantages over that o cards, which constitute virtually the only alternative at the present time. In the first place, it is more interesting—or so, at least, I have found it—both to the experimenter and to the percipients; and this Means that it is much easier to induce ordinary people to work and to continue working at the subject. In the second, the very large Tange of possible originals invests each with, so to say, an indi- a 1The Cross-references and Notes given after the Catalogue should be help- ful in this kind of situation, but it is virtually impossible to make them so €xhaustive as to cover all possible contingencies, and the serious big Sele pete Should accordingly be at pains to study the Catalogue carefully and to familiarize imself thoroughly with it. The Notes should always be consulted.

62] Proceedings of the American Society for Psychical Resear viduality of its own, which is almost wholly lacking in cards especially in the five-symbol Zener cards now commonly used, point of this is that it enables us to connect, as it were, a drawiy of a hand, say, made on occasion A with an original of a g displayed on occasion B, and to inquire, by the study of many suc cases, whether there is any non-chance relation between the events; whereas, with cards, the individual object too soon gets log so to put it, in the crowd—for who shall say whether a guess of circle is to be regarded as possibly retrocognitive on the last cire among the target cards or as possibly precognitive on the next? I do not, of course, pretend that the phenomenon of displacemen which I venture to regard as a key fact in the subject, would ney; have been made manifest with the aid of cards alone; but there es be no doubt that its discovery was greatly facilitated by this proper of drawings. Another merit arising out of this property is that it enables us | study ‘cross-influences’ between groups, as in Expt. VII whe different groups scored freely on each others’ originals, or withi groups, where recent work has shown strong evidence of ‘later telepathy’ or some analogous effect, and possibly of the formatic of a kind of ‘group mind.’ : Moreover, the fact that a drawing has both form and content, ( ‘shape’ and ‘idea’), and that different objects and drawings there may have varying relations of similarity with each other, bo invites and makes possible enquiries into the nature of the proces involved which would be out of the question with test materi lacking these properties. On the other hand, these last mentioned properties inevitab introduce, on occasion, a certain measure of ambiguity; this can on be completely eliminated by adopting a policy of literalism whie may sometimes seriously impair the sensitivity of the technique. What is wanted is clearly a technique based on a type of tes material which, while preserving some degree of human or dramat interest, shall provide a range of possible and different test-objeet comparable with that of drawings (say, not less than 500), wit complete lack of ambiguity, a theoretical expectation and variance and the possibility of “scoring casy successes as well as difficull be welcomed, for the problem of satisfying all these conditions j far from a simple one. 7 Meanwhile, the technique with drawings discussed above, pro vided it be not falsely regarded as more than it is, appears to con stitute if not a perfect at least an entirely practical instrument ¢ research ; and I earnestly hope that, in view of the great importane of these phenomena and their implications, there will not be lackiny experimenters, and particularly professional psychologists, willing tt make trial of it on an extensive scale. Experientia docet; doceat experientia. its on the Paranormal Cognition of Drawings [63 n to the many collaborators and others rn a , most sincere thanks are due to the members of the — # We mmittee, notably Professor Broad, Professor Price an ‘ brides COE their constant encouragement, advice and help. pale tle proportion of the work was done during my tenure - a ae Studentship in Psychical Research, and I am correspondingly ited for the facilities accorded me by the Electors and Super- indec yisors of that benefaction. Expervme? In additio REFERENCES. XLVI, pp. 34 ff. XLVI, pp. 277 ff. : Vol. XXXIV, pp. 181 ff. : Vol. XXXVIII, pp. 223 ff. Vol. XXXIX, pp. 375 ff. S Vol. XXXIX, pp. 266 ff. Bulletin XXIII, Boston Society for Psychic Research. , : i h, Vos Proceedings, Society for Psychical Researc tA ‘ “ “ ‘ “ a: 2. 3. 4, Ds 6. ie

64] Proceedings of the American S ociety for Psychical Researc} Appendix I Catalogue of Frequencies of Objects Drawn or Mentioned in Experiments I to VII Experiments on the Paranormal Cognition of Drawings [65 Foreword to Catalogue The left-hand column gives the names of the objects, etc., drawn or mentioned by percipients. For the differences between names printed in different sorts of type, see sections 25 and 26. The next column shows the number of the experiment, if any, in which a drawing of the object named was used as an original; e.g., a drawing of an ARROw was used as an original in Expt. V. Where the entry VI* occurs, it means that the object appeared among the forty originals drawn for potential use in Expt. VI, but not actually used. Cf. my second paper (2) p. 280. The entry VII* indicates that the name of the object was written on one of the 100 slips sent to experimenters of Expt. VII, but not used by them. Cf. (2), p. 296. Recent work shows that the percipients of VII scored significantly above chance on these objects; and experimenters should therefore treat them as if they had been actually used as originals. The same seems to be true to a lesser extent of the objects marked VI*, so that the same precaution should be observed. That is to say, N should be taken as 495 for objects marked VI* and 496 for those marked VII* and the number of entries under VI and VII respectively should be deducted from the totals in determining the value of c. The various sections of Expt. VII are distinguished by the appropriate letters (Ibid. p. 294). The next pair of columns show the number of percipients who drew or mentioned the object concerned in the course of the first five experimenters, either as the sole or virtually sole content of a draw- ing (a) or in conjunction with another object or objects (8). Cf. Section 24. Similar data for Expts. VI and VII are given in the pairs of columns indicated by these figures. The last three columns give the sum of the a entries, the sum of the 8 entries, and the sum of all entries in the line, in order. Cross-references and Notes, which I fear I cannot claim to be exhaustive, are given immediately after the Catalogue. Experimenters should make a practice of always referring to these when using the Catalogue,

66] Proceedings of the American Society for Psychical Resear ‘] ABBREVIATIONS. The following abbreviations are used: o = Approximately, more or less, substantially, etc. Carprs. = Carpenter’s. Chen. = Chemical. Convl. = Conventional. D or Drg. = Drawing. Geoml. = Geometrical. Incl. = Includes; or Including. Illd. = Illustrated. Indt. = Indeterminate. Mechl. = Mechanical. n.o.1, = Not otherwise listed. n.s.l. = Not separately listed. O or Orgl. = Original. Ord. = Ordinary. Ornl. = Ornamental. 0.t. = Other than. pr. = Presumable, presumably, etc. Sp. Mn. = Specifically mentioned. Unspd. = Unspecified. Var. = Various. of in Experiments I to VII me of OBJECT Orgl.in I-V VI * Expt. -a 9p i) 18 sacus or BALL-FRAME Sate 1 Ace of Clubs - Sune ave wae RR a a Me is ne er ea AIRSHIP (Dirigible) - + 2 4 : eros, Sex Crocoaile ae Aa (Christian) - - - ; : Ancestor +, 3° Se eee ; : Ancor <b ae te 1 tat t RII ie are a ere a Indt. or Unspd. eh, ss j $s s Uta ELOPE of ~ a) gee ‘ ety a 19 9 24 14 H or ARCHWAY, Ord. - 4°53 ra (Arc de Triommme - i v4 , Noah’s - - : Z , Sp. Mn., etc. - - Pam A at | our = - ; 1 or. See Troops (Human EINGS) - . Arrow, Full j= tt Pret a realy Von 4 9 Me Partial - - - - : se Re 2 cum thcan, See Dustbin SHTRAY = - ~~ ee ee ai ee OMOBILE, See Moror Car ue, See Tree ; ~ Battle = es io 1 ~ Executioner’s - - gr cul 1 —~ Woodman’s ees: 4 ‘ Baby, See Human BEINGS Bac, Hand - - at RR Oe ee Travelling - - CaN ea athe ‘ ANCE a Scales, Ord. SEV “i 1 pring in! deen ae . 5 . Indt., or var. - - - Be Jeb oad eet. ale eee i | CE Football, Assn. - - ; [Oty Ee Rugby - - Bisel aaa SSE erat 4 ; Z Pe A ee ee 1 BE Woon. es ha Nh ed _* colored, heap of - - ; “her Catalogue of Frequencies Objects Drawn or Mentioned VII a £p Te 7 4 os ie Le ese eae ane ee 19 20 3 A Ps ae Pes Binds 2 “aay wer pe Ae aio aes 7 area Sia “eae ee -iments on the Paranormal Cognition of Drawings meprper [67 TOTALS a £p Total Linas 1 oh ie 1 33°..°6 41 Be 3 3 2 2 1 1 1.2 3 aun 1 17. 4 21 sisi 1 5.23 8 13h ut 17 Boats 2 | & 62 43 105 11 8 19 Bila 2 oe 3 $4 3 4 ae | 1 12/14 26 os 10 4 4 eS" 1 Ese 2 ee 4 aaa | 1 Tine 9 anes Z sea 1 ALS 11 hae 1 16 21 37 Lone 4 4 1 5 Si -12 7 song 2 1 ee 9 ome 4 we 1

Expt ia). 8 Orgl. in Battoon, Free, Gondola of 4 Kite or Barrage Child’s - = Battoons, Child’s, Bunch of Balustrade - . « Bamboos or ~ - BANANA - = Bandstand . ~- Banisters, Sp. Mn. BANTO Qore oo tc Barge, Stern only BaARoMETER - = Barret or Cask es Water built - Barrow, ~ Coster’s - ff ~ Porter’s Bastn, See Bow. Basket, Empty or unspd. 4! with contents Bat (Animal) - ~ Bar (Baseball) “ (Cricket) Batu, Ord, - Tear aot, - Battledore - Battlements - Bauble, Jesters Bayonet - Beach, Sp. Mn, - Beacon, Belisha - Bear, Ord. or unspd. “ ‘Teddy’ R VI* * NED: er Sew ame gS i ca | fu Bee A aE 0 ere pe ee sg rR Se EN ee ae moe a Ben 7Ord, 2 Bree) o> fee BEEHIVE - BeEtig~ +. 3 Bett, ~ Church : Hand Desk Bell-Push - BELLows - Bett, Mechl. - trousers © <\\ Ge BENCH, ~ Garden Seat - BECVCERM a yi tic AS BINOQCULARS (51) fC), a2 Brrp, single, ord. or unspd., a SOE none Roe eS as above, flying - - plural, ord. or unspd., flying or unspd. - fantastic, toy - - young, in nest “Chick - “Cock - Cockatoo Crow - “Ne HINGES iS. hrs patel a cs VIIx VII* Sa IT* BO FE re; eo ee BERR, ae Nad 1 2 Se ey Cmte pei ee p= greys Vv : VI*¥ 6 V ce Soa BS ak eg et ks S23 = Oe eee cs ea OR ie SR ae a @ dno I-V a 1 2 -_an’ 8 VI mide 2 oe a 8 VII 6 1 12 STG Sak re ® Ge Sa ce ys mepe i * Mtoe. 4 [e) “Os RH ORNNROF 11 mR ORN.- med. - o-- ioe Oe alt ares eo Ex, se Eagle Goose Gull - en - Kiwi - Ostrich Owl - Parrakeet Parrot - Peacock - Pelican - Penguin - (iE Dee RL See ee OR: et cece pe a Poultry, unspd. Robin Spe BeoOR = =| Swan - = Toucan - Turkey - Vulture - cay = Q Ss oa RE PS S 1 Sal abe De ks, Child’s toy lotter, Curved, hand _“ Flat, desk - Slg., small, pl. m (not Reel) » Steam - Si e(s) o.t. skeleton NFIR 90K, single or “4 Ook-ends ee ‘okmarker - Name. of Osyecr Pigeon or Dove otor or Speed Bowing - - i i« Racing eight Small, open, indt. - Indt., forepart only x Ss & S be ~S <i = by er ee se n(8), Ord. or Ridin G: “ “aes Var, & unspd. ilk or ~ - ine or ~ - Rubber, Hot-water Swallow or Swift OAT, Sig., F & A, 1 mast =) slg, F & and 2 masts Sig., Sq. rig, 1 mast “Mast but no Sail - Sail only - - g Orgl. in ' 6 et a IO 2 SE ed et Expt. Vil* se VII* Vv VII* VI IV I-V 0. 2 2 1 2 + me De ue J * NOR. » YENE D _. - ee pe: Or: : a VI a £p eet Ob op Via 1 4 1 b 10%.) j reat 1 : a Nig ieee ae 11 4 py 2 ies | er hess t% 1 a Pe Carias Y Dieaeta | 2) 4 to ie 3 Bee eS om | Bo ik Das 1 F O83", 11 O72 ; 1 1 : : 1 bite Be ee 02 5 oe 6.4 a » Bw NON e mee to urs - 7 wR D- : periments on the Paranormal Cognition of Drawings Vil B eee Oe — = WWW: + [69 TOTALS Bp Total _ NEN ANKOURUARAW mn: — CO » BND. OFM KPNNAOCH- »* wWwNe i) en OMe ee) RB DON: HB RENWOW: ee SO Ie ms CDN: _ =O | all © On et ee et TD IT = 0 Gn DO iS) VE PENNE SUR ONO wo NRE DD Nee ewe iy — n> QHAH ADRESS Noe mm GO COR

78] Proceedings of Name of Oxsjecr Hockey Stick Hoe (Garden) Hoox, Fish - Nig Meat - - Plain Hoop, Child’s Horn, Drinking, on stand if Motor, See Moror Horn Horns, Animal, unspd. Horsz, Ord.- - - “Model or Toy “Rocking - “Mask of mene HorsesHor - - Hose, Fire or ~ - Hour-ciass or Ecc-TIMER House or Corrace, Ord. o> the Onspde: e238 “Public or “oy - Howdah - - Hub only of Wheel - Human BEIncs: Man, Youth, Boy, var. £ Head only “ “ “ (Comic) “Archer - Aviator - “Bishop - Cannibal Chinaman Clergyman Clown - Cricketer “ Cyclist = - Diver, Deep sea Dragoon - - Fireman - Fisherman (Seafar- ing) Fisherman (Angler) “Footballer - General - pale S Pr s2 1, Sanoae ater “Indian (East) “ “ (Red) Jester - - Jockey - - Ot RCC IEE 1 = ae fe. ‘eerie Native, unspd. Negro - - Organ-grinder Parachutist - Piper tilieaaa 2 Policeman - wx) Sailor CRO) is < = the American Society for Psychical Rese gxperit Ce ee ee TN as ne ve _ RO: + ORR gy. - me oe pmo RN. « Sos deers : he. A Humpty Dumpty - © fabs 8 ew oe Be PRE yne oh peas o bee pyand, Sp. Mn, - Be ~ Sphi, with tid E ‘ ni i ee ae NE ee ee Pek Oe Gh eee a oe ce Se | Schoolmaster or Don Name of Opyect Sentry bi is : Septet - Skater (Ice or Roller): = Wertrae Snake Charmer - Soldier, Indt. - Student or Scholar Troops, unspd. Tweedledum & widte Undergraduate - iking - ties Air raid Woman or Girl, var. - “ Head only - Brittania or ~ Chinese - Dancer (Ballet, etc.) - = Dutch Egyptian Madonna Nude - Nun - Nurse, Child’s : Hospital Red Riding Hood Refugee - - Witch - Couple (Man & Woman) - Children, Baby == = Child, M or unspd. - - i Child, Fem. - 4 Two. M & Pl., var. _Crowd, ‘People, etc. ae ee ee et or Shed, var. - Ic “Cream Cart (Tricycle) come - - h pity Si ion. ’ =. or WELL ect, Indt. = lator, Electric » Domestic - : k-in-the-Box be ma —~Jlam - - “Gor Ewer - i BEABOO! 50 Suk cae i NEL, Dog a ea ~ nents on the Paranormal Cognition of Drawings »~ Ne pe wom D a B- WNWORF Ae NKWHWWHEN Are - ee wo » YNORe no- WERENENAPENUE HNAREE ov * etwhrds + °¢ + RONRR: ry _ oO Rt Mee wD NN: ENO &: ON: / YD: * me D ie Be tek: an) oz, er Seek Pe Te ope et 6 eee fk epee slp Ne ee ee NR ROE -_. ry an = Rs woo Fee Pin Sen pene Mak BS WeNs ww Nw . ee oe eee OW Ne Recipe: “Sag IN Ms eee NX. Fw. Ue WP _ COrPNNOO _ Ne Mand p> __ _. - (ieee. cht heat Tt ae Ala Set | . -_ . PRR OrRRNN.- d KK DPR RPORrENNE tog Nae wae one ee NRF oO- =" ore Ure wn

80] Proceedings Name of Osyjecr US ova 3 Samak een Rey Ord. (eave wate “Yale or special Kitchen Range - - Kite - os Knire, Paper - - - Pen or Pocket it Table - = Knot (in rope) - - Danner sion. opie Uli ADU RT eis at hat Nal Sie Lake of Sh GGh Wim es = JUROR en eee eT ci Lamp, Cycle - - - Electric (See note) Floor, tall - - Hanging, unspd. - 13 Oil, Heating - - Spirit, glass - - DIGeE? eat a Dable vars") 2 Lampshade (No lamp) - Lance, See Sprar Lantern, Ord., Hand or Lead or Leash, Dog - Lear, single, var. - “=~ 2 or 3 or few - BL, Mimspay te Ligcterig cis Se Leek - hanging - - “~ Projection - Tarieng = heap rey nc Launcu i ae Lawn, Sp. Mn. - - - “ec iS) os SP eee ee oh Legs of Man, See Triskelion Lerrer, Postal, See ENVELOPE Letter Rack a ee LIFEBELT ee LicHtTHoUSE - Lightning - .- Linx, Chain or ~ “ Cuff ms my Eton, Ord: =". Heraldic or ~ Pizanp 40 ER cs Loar et Geek Logster or CrAYFIs Lock, o.t., Padlock DROP EE ke ee Locomotive, Ord., no train “ ““ with “ 4 Child’s Loop of String, etc. - - Luggage, Unspd. yams Lyre Lia, RR, Tones eat PO A RE Fe PE. oe Bey Machinery, unspd. - - Macniryinc Grass - - ~ j wooden - Orgl.in I-V Expt. “he 8 VE) ed Wiie 1004 : ae Vilt* 4 1 ; “oe : ee Vie, 4 ‘ ad VIla 10 3 : ae ae eae ae os 4 : eae ee ae : ae : as | : yO, Vin wie Ae June i oad fa a 7 2 Ae | ee Vie 4 3 : be hey : Bes : mite 2 Rae ery j eit Vit. 404 ; 7 ae : ey Mie ar. : eas eee Ae ie bse 1 1 ‘OBR DOR R Uk ee + eee VI B * ORR. the American Society for Psychical Resea VII _ * NebD- Com Re me OO Rg = NASH. . ea a 1 ee 10 B E periments on the Paranormal Cognition of Drawings us TOR Orgl.in I-V VI be L rN 0 Name. of OBJECT Exzpu 90 2 B ie Carpenter’s- - - 1 * a See HuMAN ssc al f 1 i olin See Wereoee SN 6 4 a ‘ By, Billiards = =) > = ; pa ae 6 Baspow, Vee» OF pr. - = ; ne "te 13 Mee, Ord. or plam = -- = : bart ae Bees . Se ee Aen oe! |e: % i oree, See Horse : f Mat, ornl., Lae 2 * Bees ; 3 3 MarcH, not in box - - i Meadow, See Field Gs i Ee ‘ ; me. 2 by ah Ve oe ; eduso a é i is - . . . 1 fecaPHONE - - - = 7 vie Nie ; 2 ae. - Dee : ct 1 a 5 nk DAC ne ie ee ye ee “ . = oe 1 epg Milk ‘Churn’or Can - - - or i Mae WMS keel of Pas 1 iPass ee 5 a a “ wheel of cet D3 sere oe Mistletoe - = : ei "Sia RE (Bishop's) cote (ame a Ma del (Artist's dummy - : - fi a KEY or APE SiS en LE ; : ; ; nF 22 Moon, Crescent - - - - Z rata ; 13 m6 Full or pr. - - = 5 : ; —° Maninthe-- - - P gee mas : ; Moor or Moorland, Sp. mn. - . i i 1 ee ew Ne aoe i : ‘ i = ee ne 7 Mororcar, : 1 _ “Side view or unspd. l VIIe 4 : 4 : : ' 1 “Front or rear view § ' : nae 13 13° Motor Ambulance - - - é 5 1 @ {0T0R Bus, See Bus. Te 2 4 MOTOR Cycie, solo <j i= ° « ; : iS < i 1 = ..“ with Sidecar _- Meme ota 7 a Mo oo (Bulb) - - VIlb . Nay ONC 1 mee, Sh. Mn.- - = +» E bipR ts : : : meowtiain or Hill, Ord. yy Ni 7 : . : ¥ 17. 10 . “ Notably peaked § ; aes 4 4 j “ side, Sp. Mn. - 7 " 5 : ee 14 BeUseor Rar - - - - VII* 3 . : ' a BAT AD oe fens ese ek 22 im tache, no face - - - IV sions Pe an 2 1g Muc or Dancin 4 Ankate 4 ' ; 1 pee? ; i : YSHROOM or ToapsTooL - 41 ‘ 60s 14 a Paaeg Cais 1 aig ida Ma) i ey Paar 2 [81 TOTALS a ¢ Total 4 3 1 1 ee a3 Nal | 1 seo 4 Aas 4 dass 1 3. 3 1 1 3 2 aoe 1 Tas 1 aed 1 2 a 3 Nee pea | Nas : 2 i! Dees 1 1% 2 s Nae 2 ee 2 5 4 2 > ae 1 <r! 4 Mas, 1 Bio. 40 9 9 18 : a 8 Bat ae ie a2 2 ch ng 2 iI 2 19 5 24 4 4 8 ial ae 2 pean" | a Denes 1 oy 1 ae 2 10 69 79 10 13 23 pene) 9 12 4 416 bathe 4 je tee 15 5 20 ec ines. 20.4 . 2h ene 1 Bors. wld

84] Proceedings of the American Society for Psychical Rese TO Name of Osjecr Pyramip, Ord., rectlr. Other shapes “ Quay or Wharf - - Question Mark - - Quotation or Inscription RaBBIT - - - - Racket, Tennis - - Radish - - - «+ Railings, See Fence Rattway or Tram LInEs Rain, Sp. Mn. - - RaAInBow ah Rake, Ord. - - Farm, wheded Rattle, Chiiks. a, “Gas or Police Rays, of Sun, Sp. Mn. Razor, Safety - - Reeds or Bulrushes - REEL, Cotton, etc. - “Fishing, etc. - Retort, Cheml. - - Retort-stand - - Revotver, See Pistou RHINOCEROS - - - Ribbon, unspd. - - Rirrtz, See Gun Rinc, Annulus only - “ Road - Rock(s), Ord, Sp. ie’. - “ High, with aes | Rocker - Rotter, Gate, e Indt. re Steam Roller Press Rottinc Pin” - ee 8 Roof(s) only of House(s) Rosette, Election cae Roundabout (Merry-go-round Rubber, India, See risiypicil Rucksack - Rug, Floor or Hint “Travelling - Ruter (Geoml.) - Sazor or Coc Sack or~ - SADDLE - - Safe ues SaFety PIN - Sail, See Boat rectlr., Mexican esto) eee hi ae CSE 4 Ee Oe eg og Finger, with stone bs Signet - Or UNGER, Sree fe! oe River, Sp. Mus =) = = Nad: aie yen cies es | Orgl. in Expt. VLE Vi VII* VIIc VII* VI* VI* VIld /_ RD PER D he Ny RoE —= be ee Nt ea N ry epo. ee COR ep mB DO BH. -_ [RR RONEN wT er rey er. + = Vhb- © A pmb es: BMrRONM: Seon. . = Re eee ft ete 8 =e RRMDN Dm. -. NeeENwW w- Experiments on the ame. of OxByECT m-porse =~ ; ecrow i. ‘z Paper, etc. - Stonework - Meee - - - = bp. Mn. - - ma Anemone - - Merforseé - - + (Animal) - - Cbs 0 0 EE eos 3 sheath, Cyl., for Hook - (Artillery) - - (Conchological) - i Air sag - age. or sneha. Sailing, Ord. - ~ Galleon Steam, Ord. - - B. Ware sesh HOE, Faced. pe M. = bedroom slipper Mee Tree . . a Incident - “VEL, Ord., See SPADE hawl s es = HEAF of Wheat, Corn, etc. Pp or RAM, single or ~ «“ « "pl. or flock -., ord., high- heeled or walking - ild’s ae. H00ting, Word (See: note) for Incendia bomb Mhitedded Wheat >. Paranormal Cognition of Drawings Orgl. in Expt. Ill vIl* IV Vil* Vila Vil* vie VIIb vI* II - De * PR we + -_ Ome: +s ep - Ord . De Ree: al ao | . ROOF: + oe Or. et dS . =e OMe Ces Bete - . hee ee RR eee eae Pe ee . ure « We - moe Rr &:: D- a ee «whee ra Pe wr C+ Ms POR: HOPPE PH: R&R: Ne: N+ — Ne NDAD LS) Nwua _ nN _ =p Orrerathre au _ me ORR Urs Ws De: > ae Roe N&re: = wo. — Oe« oa Re HB prewhdr: e. mom eed at he »e NHKWMDO _ =e —e- [85 TOTALS gp Total 2 9 1 6 = 1 2 6 50 4 7 2 1 2 1 1 2 nd 8 1 2 3 1 1 5 1 2 8 1 2 2 3 1 1 1 4 6 1 9 9 2 ee 3 2 3 9 LS) 4 6 2 > 5 1 22 ry — Re NEO

; 90] Proceedings Name of Osject Viscera, Indt. - VoLcANo aot Nimibae Wall oe Wand, ~ Fairy’s WARDROBE or ~ - WRB: Mine jee) ie WatcH - of the American Society for Psychical Resea VI Water, Ser ek Unspd, Sp. “ Peg of, fro hese Water-Burt, See BARREL WATERFALL - - ~ WateRING CAN - - Wave or Ripple, Sb. Mn. Weathercock i Weir es oa Well-Head - - Warts ors 5 Wharf, See Quay Wueet, Ord., ~ Cart Pin wicar "Steering Ship’s Steering “ Toothed ~ WHEELBARROW - - Wines 0. Gey 2 WHISTLE i u wan (Stumps) - ( Wiecwam het Wind, Sp. Mn. - Winpmit - - Winnow, — Church, ies Domestic - “ Sho hop Wineglass (Possibly Egg-cu Ni Wings, Unspd. - Winning Post - Wrretess Ser - Wishbone - - cfs: 8 be Perego Age SY pe Mane ss Ae ee Eos s AS CE ee vind te ta ee ae eee ety Woman, See Human a Wood or Forest, See TREE Woot, See BALL - Wane wie Over alen a Wreath - - Wrencu, See Srawner Wrincer or MANGLE - Zreee) cee) Zipp FASTENER or ZrrPER Orgl. in Expt. Vild Vile VII* VII* VII* VII* VIId VI* L-V a £ 3 1-96 “ae Aas 4 6 ae ee eat ae | Fist aaa | eeu, Pos, 6 1 ed oe fui; en ee 22 Waicie en si 3 hee 2 1 a 1 2 ore a me ore ee OD RD: po r- " mb. Wr. B a '* CNWHN. VII a £B 3 2 t oe ae ae 4 2 1 2° 4 eae ad se a 6 ae 4:8 aie ae Wee a il 1a 6.3 5 19 OY ae So pe 4 2 TOT a 6 1 8 3 2a 4 1 a 8 a 2. 16m 6 «of 16 a 3. 2 2a . if 4 33 2 = 18 +7 1 3 so 2 = 15 . ae lS 1 3 3° 1 Ge 31 Om 5 «-. 9 34 1.3 4 ws 13° 4 3 1) & 3a 13 : 2 2 aaa 2 =z Experiments on the Paranormal Cognition of Drawings [91 Catalogue I - VII: Summary Total Headings 556 Entries, (1 aly — Total Headings 703 «Entries, Mh 8 “ Bs a+: “ “ Total Headings 1,259 “Entries, es ‘ “ Bs: a+p: Total Headings 114 “ Entries, Ne “ “ B - “ “ a+p: Total Headings 1,145 . Entries iia “ce “ 8 a+B: “c “ Ratio a/p Number of percipients Items per percipient Indeterminate Geometrical Letters of Alphabet Numbers or Numerals Total Gross Granp ToraL I-V VI VII TOTALS a. B a a a 8B A. RoMAN type items. 1,209 1,430 1,244 3,883 5 70. 897 2,186 1,795 2,133 2,141 6, 6,069 B. Italic type items. 547 473 333 1,353 649 598 626 1,873 1,196 1,071 959 3,226 C. Roman and Italic combined. 1,756 1,903 1,577 5,236 1,235 1,301 1523 4,059 2,991 3,204 3,100 9,295 D. Illegitimate or Doubtful. 108 69 42 219 243 186 184 613 351 255 226 832 E. Net Relevant Items. 1,648 1,834 1,535 5,017 992 1,115 1,339 3,446 2,640 2,949 2,874 8,463 1.66 1.64 1.15 1.46 250 246 245 741 10.56 11.99 11.73 11.42 F. Indeterminate, Geometrical, etc. 292 78 21 391 207 33 30 270 25 7 3 35 11 4 0 15 535 122 54 bal 3,526 3,326 3,154 ~ 10,006

94] Proceedings of the American Society for Psychical Resea Cross, St. Andrews: Does not include ‘multiplication signs’. Cross, ‘plus-sign’: does not include small crosses incidental to churches, — CrossBonE: Bone. Cross-stitcH: Only included because O for I. Only plausible candi¢ are mentions of Embroidery and of Stitch, qv. Crown: Wreath. Includes Diadem and Coronet; also two of Laurel, specifi described as ‘Crown’. Cruer: Pepperpot; Salt Sprinkler. CrystaL: Stone, precious. Cup-and-Saucer: In view of the fact that the O for VIld was a Cup w th a saucer, it seems preferable to treat cups standing on saucers as 4 stituting single entities. ji Currain: Includes one or two instances of Drapery not on figures, Cusuton: Pincushion; Pouffe. DaccEr: Bayonet; Sword. DartzoarD: Target. Decanter: Bottle; Carafe; Flask. Devit: Angel. Disu, Shallow: Plate. diameter: similarly for following entry. DitcH: Trench. Diviwers: Scissors. Dock: Pier; Quay. Doe: Includes a few cases of Hounds and one stuffed dog. Drawinc Boarp: Pastry board. Dress: Crinoline. Includes Skirt. Ear: Not parts of faces even if sp. mn, Ecc: Not mere ovals or ellipses, which would be excluded as geoml. Ecc-cup: Wineglass. Emprorery: Cross-stitch. EnveLope: Not mere square or rectangle with diagonals. Exrourate: A bad O for II, illustrated by a drawing of one large leaf two small. Eye: See Ear. FreatHer: Pen (Quill). Fence: The classification is roughly according to whether the main lines 2 vertically or horizontally. Fess: A fess is a band drawn horizontally across an escutcheon or shi and an embattled fess is one with the top indented as by battle: The O for I naturally showed both Shield and Fess, but there is no st drawing. The nearest are listed under Battlements, showing these but shield. Fincer: Hand; Thumb. Fis: Haddock; Whale. FrrepLace: Kitchen Range; Bonfire, Fisu Sutce: Trowel. Frac: The O for IV showed a Black Flag bearing a very conspicuous Cross. Flags merely incidental to Boats, etc., not included. FracstarFr: Not small staffs or poles incidental to Flags. Fiask: Carafe. Fieur-de-tys: Halberd. Fiower: Plant. Those listed as Indt. or unspd. comprise all unidentifia flowers, whether single or plural, in vases, etc., or not. Considerable c has been taken, with external assistance, over the identification of t classified flowers; but botanical accuracy cannot be guaranteed. Fry: Dragonfly; Wasp. Foor, Winged: Angel. The connection with Mercury as messenger is strikit Fork, Garden: Trident. Fountain: Geyser; Water, jet of. | ~Expe "FROG: Toad. GALLO peysER: Fountain. Guove: Hand. GOBLET: Brim. Gop: Angel. GoNnDOLA " Beet : Coat. j ; Cor MOPHONE : Musical Box. Grapes: Balloons, bunch of. Grave: Coffin; Corpses. = 1/5 deep: Bowl. ~ 1/5 deep means of depth about 1/5 of iments on the Paranormal Cognition of Drawings ws: The O of VI* was actually drawn as a Gibbet. ad Door; Turnstile. E: Portcullis; Trellis. a Shooting ; Pistol. GyroscoPe: Spinning Top. ERD: Axe. ee Chessmen; Pawn. Hammer: Mallet. Hanp: Glove; Finger; Thumb. Be oe Col The Hat drawn as O for VI* appears to be a soft fe i i feminine. The Hats i line appearance, but might be ema jnctude two Bonnets (Expts. VI and VII) anc display of Hats in a Milliner’s window (VIDS Heappress, Red Indian: Shows no Indian. Hence: Fence. ; Hewrx: Spring; Spiral. HenceHnoG: Porcupine. a : Almost certainly determined in VI by the Orgl. Bow (TIE). and N as 5 and 495. i Bocce: Building; Doll’s House; Hut; Roof > Street. toca Human Betnes: It is impossible to deal with win a pi weg te eet chy Te separately onl would be of little interest. av ( ee i tant, interesting or conspicuous Vv. ae por té cae figures, but any may be plural. Some of the special are drawn as ‘head only’. Humpry Dumpty: Egg. Hur: House, ete. Ictoo: Hut. ; Inxpor: Includes Bottle of Ink, n.s.l. as Bottle. Insutator: As on telegraph pole. Kitcuen Rance: Cooker; Fireplace. Krre: Includes conventional Kite shapes, tail, etc. : Knire, Table: Includes Bread Knives. Knor: Bow; Noose. Lapper: Steps, household. Lape: Spoon. poze: Pond. AMB: Sheep. ? Lamp: Shades are not separately listed when on O for VI was an old-fashioned oil table-lamp w Lear: Cf. Note on Exfoliate. ae OCoMOTIVE: Tractor. Locomotives hits on the O; I have separated them from the o of possible interest. even if not shown with Lamps. The Lamp dr: ith no shade. trains are of course elig thers merely as a

98] Proceedings of the American Society for Psychical Resea Sprnninc Top: Gyroscope. SprraL: Spring; Swirl. N.B. These are flat Spirals, not Helices. Spire; These are cases in which the rest of the Church is not shown. Spoon: Ladle. Sprine: Helix. Starrs: Steps. a Sranparp: Illustrated for II by a Flag with a Lion Rampant. The ¢ bination has not yet been drawn. [ Star: Solomon’s Seal: The 11 instances of I-V are probably best ignoreg ‘geometrical’ items excluded by later instructions. Statue: Includes Busts. Steps: Stairs. Srop-cock: Water, Jet of. Srop-sicn: Beacon; Policeman; Signal; Traffic Light; Winning Post. cludes (VI) Hand with word ‘Stop.’ iy Strap: Belt. Stream: Canal; River. Srreet: House (pl.) Strives: Tiger; Zebra. F Sun: Only when definitely intended for Sun; eg., circles with radiating |; lacking description or context, are not counted. Swirt: Spiral; Splash. Sworp: Dagger. SyitpH: Angel, etc. TABLECLOTH: Counterpane, Rug, etc. Tare Measure: Ruler; Scale. Tarcer: Not mere concentric circles. Tent: Wigwam. THorN: Needle; Pin. THRONE: Chair, ceremonial. TuHums: Finger; Hand. Tie: Bow; Knot. Ties: Chessboard. Toap: Frog. TomBstonE: Coffin; Grave; Milestone. Tones: Pincers. ToweEL: Scarf, etc. Tower: Chimney; Lighthouse; Pylon. Trarric Licut: Stop-sign, etc. Tramcar: Bus. Tray, Household: Includes a flat Meat Tin. im Tree: Branch; Stump; Trunk; Wood. The O for V represented a con tional and somewhat ‘cabbagy’ Tree, which might have been Chestnut, C or Sycamore, but not any of the others; I have bracketed the first { classes accordingly. : Trerort: Ace of Clubs. Tretuis: Grille; Portcullis. TRENCH: Ditch. Tricycte: Bicycle; Ice-cream cart. Tripent: Fork. Trousers: Pants. Trumpet: Bugle; Clarionet; Horn; Saxophone. Includes two coiled Hi (VII) described as Trumpets and three ~ straight or ‘heraldic’ tr me as well as one true Trumpet (VI). } Tuse: Hose; Pipe. Turnip: Onion. Vase: The Amphorae of I-V may have been partly determined by the EM or Juc used as an Orgl. in V. Take c and N as 4 and 491 for safety. — ) Experiments on the Paranormal Cognition of Drawings [99 Bae panie P ireadetlit etc, Includes Cello, Double Bass, Viola, etc., if any. Wat: Not of Houses, etc. WASP: es Se : Clock. ; abe Outdoor: Lake; Pond; River; Sea; Stream. Water, Jet of: Fountain ; Stop-cock. WATERFALL: Weir. : va ‘ i lines or incidental waves on Sea. _ Va if eeedy incidental to Church; only if Sp. Mn. or v. consp. : il. . . . Vere in VI has beam and counterpoise instead of winch. WueeEL: Spinning Wheel. WHEELBARROW: Barrow. WicwaM: Tent. : Mill. : ( Mba Not mere rectangles with cross lines. Winpow, Shop: Stall. j Wineglasses but : Ege-cup. The first class were judged to be vo aeuiaty ro Egg-cups; the second class, by shape or context could not be. } nee Wrnninc Post: Stop-sign; Traffic Light, etc. Wrretess Ser: Gramophone. Worm: Caterpillar; Snake. WreaTH: Crown.

Experiments on the Paranormal Cognition of Drawings [101 Appendix IT. b UNIT SCORES FOR N = 491 100] Proceedings of the American Society for Psychical Resear Appendix II. a Hits Misses Hits Misses UNIT SCORES for N = 741 c U Vce Ug! c c U Vee Uy c Hits i ‘ a 0 2687 0147 +.00547—s 1 25 1210 .0816 .06748 26 oy a wei Misse 1 2.381 0231 «100972:««2:S; «So 26Ss«d:91S 0824691827 Neh oe te Mae a eas Ne 2 2 NS ae a ee ae oe) ee 3 0352 99 4 28 «11 0839 . pgrl : ; peers 1 53 1.055 0873 .08282 4 1973 0400 02025 5 29 1.141 0845 += .07408 30 a eat Oa eee 2 54 1.046 0876 —.08376 5 1990 .0441 02334 6 30. 1125 .0851 07565 31 eee eet ee 57 1.020 0883 08653 6 1.821 .0479 02628 7 33 1.080 .0866 08020 34 FR ic aire Se 58 1.012 0885 08743 7 1.760 .0512 02909 8 34 1.066 0870» 08166 35 A eet ee ee 59 1.004 0887 —_.08832 8 1706 .0542 03179 9 36 1.039 + .0878 += 08451 37 a yaa I a 63 972 0892 .09179 9 1.658 .0570 .03439 10 37 1.026 0881 .08590 38 Fg eee Iie 64 964 .0893 09264 10 1.614 .0596 03690 11 49 992 .0900 10091 50 bce eae eee 65 957 0895 09348 11 1.573 0619-03934 12 50 ©.882.—=s««.0900-Ss««10204—S ‘5 Ro ENE os A 66 950 .0896 .09430 12 1.536 .0640 04170 13 55 836 © .0898-~—Sé«s10742Ss«56 ae ee eae of 943 0897-09512 13. 1.502 .0661 04399 14 56 828. += .0898.-~S «1084557 ib Mahle Ales a8 2 908 0900 —_.09908 14 1470 0679 04622 15 ee | eee ea 2 RY Cats ae eee 73 901 0900 + .09985 15 1.440 0697 04839 16 72 706 0868 12296 73 ei Race eles I 6 882 =—.0900~—-.10210 16 1411 .0713 05050 17 89 603 0814 13496 90 tego ae 77 875 0900 -~—-.10284 17 1.384 ©0727. 05256 18 90 597 .0809 13557 91 So ee ee ae 78 869 0900 10357 18 1359 .0742 05457 19 91 592 0806 13617 92 diay ee ees 9 863 .0900 10429 19 1.334 .0754 .05654 20 92. 587 .0803 13676 «(93 Pee eis ogee ee A 80.857 0900-10500 20 1.311 0767 05847 21 93 582 .0799 13735 94 eg eee 8 851 0900 —-.10572 21 1289 0778 06035 22 94 (5876 .0794 13792 95 RNS a aie 82 845 0899 10641 22 1.268 0789 + ©—«.06219--23 119 463 0692 =.14946 120 A A I a a 83 839 0899 ~— 107/11 23 «1.248 0798 06399 24 120 1459 0688 14982 121 bo ‘ 84 833 0898 _—-.10780 24 «1.228 «©0807» «0657625 1.483 0672 0453122 85 827. 0897-10847 22 1402 0684 04677 23 86 822 0897-10915 23 14420695 04820 24 95 .773 0888 .11490 Appendix II. c 24 1423 0706 4961 25 96 768 0887 11551 Pp : 25 1.404 0716 05100 26 97.763 += .0886.— 11612 UNIT SCORES for N = 495 26 387 0726 0523627 100 748 += .0882—s—.11789 os ee pee ells . Of ee 0880 “11847 Hits Misses Hits Misses 29 1337 0753 05633 «30 «=«103.,-—=Ss 7340878 “L901 i! Tea eae. Ie: Ray ig) Oi Tees: 30 1321 07610576131 104 729» 0876 ~—-.12017 0 2690 0146 .00543 1 21 1.293 ©0776 += 0600222 31 1.306 0769 05888 32 105 724 += 0874-12073 1 2385 .0232 .00965 2 29 1145 0844 .07369 30 S2 1292 0777-06012 33 106 720 0873 112128 2 2205 . 0206 0182 3 30 1.129 0850 07526 31 ne 278 0784 06135 34 107 «715. 0871_~—— 12183 3 2077 le (ies 4 35 1.057 +0874 ~— 08268 (36 34 1.264 0791 06256. 35 108 711 ~=—-.0870~—.12237 4 1976 .0398 02103 5 36 1.043 0877 08409 «37 351.250.0797 06376 36 114 684.0858) 112549 5 1804 .0439 02319 6 48 906 .0900 .09931 49 36 1.237 0803 06493 37-115. .680-— 0857 —_12599 6 1824 0476 02611 7 49 806 0900 10045 50 . 1s 0810 06610 38 118 667 .0850 —-.12748 7 1.763 0510 02891 8 50.886. 0900 «10158 51 ee a 0815 06724 39 119 663 0848 12796 8 1710 .0540 03160 9 6“ 72 3 18 a 39 12 0820 06837 40 120 659 0846 ~—-.12844 9 1661 .0568 .03419 10 67 7a ORR CCS 40 1488 0826 06949 41 121 655 0844 "12801 10 1617 .0593 .03668 11 92 591 0806 ~=-.13631 93 41 1.4177 0831 07060 42 141 581 = 0798-13743 11 1577 0616 0391012 93 585 0801 13691 94 42 1165 08350716843 142, 577, 0795113782 12 1540 .0638 04145 13 94 [580 0797 13748 95 43 1.154 0840 07275 44 162 514. .0743~—.14459 13. 1506 0658 0437314 95 575 0794 13804 96 44 1143 08440738245 163 511 0740 ~—-.14490 14 1.473 0677,» «0459515 96 570 0790 13860 97 45 1133 084807487 46 170 491 0721 ~—- 114689 15 1.443 .0694 04811 16 97 3565 «= .0786.—S «13916 «98 #1. 0852 07590 47 171 .488-~=—.0718—S 414717 16 1415 O7¥O. 05021) 17 98 560 .0732 13970 99 1.112 0855 07693 48 213 384 0508 15574 17 1.388 ©0725 ~=—-«.05226 18 99 555 .0778 14022 100 18 1362 .0739 05427 19 107 518 .0747 ~=—.14417 108 19 | 1.338. °.0752 .. 05623 20 108 513 ©0742 ~=— 14462109 20 1.315 0765 05814 21 48 1.102 .0859 07794 49 214 (392 49 1.092 0862 07803 50 299 234 rts “12008 30 50 1.083 .0866 07992 51 300 .233 0370 158904 30

a i 102] Proceedings of the American Society for Psychical Resear og Appendix II. d UNIT SCORES FOR N = 496 Hits Misses Hits Misse; c U Vce —U' c c U Vee —U' : 0 2691 0146 00543 1 26 «1196 ~=—-.08 4 196 0822 06872 1 2.385 0230 009642 27 1.179 ~—-.0830-~S_.07038 2 2 0296 .01340 3 28 1.162 0837 07200 3 2.077 0350 016854 29 1.146 0843-07360 41 0397 020095 30 1.130 ©0849 ~—_ 07516 5 1.895 0439 02316 «6 31 1.115 0855 07670 6 1. 0476 026087 32 1.100 0860 07821 1.764 0509 02887 8 33 1.085 0865 07969 8 1711 .0540 03155 9 34 1.071 0869 -~—s 08115 9 1.662 0567. 03413 10 35 1.058 .0874 08258 10 1618 .0593 03663 11 36 1.044 0877 08399 11 1578 0516 03905 12 371.031 0880 ~—_08537 12 1541 0638 04139 13 38 1.019 0882 08673 13 1.507 .0658 04367 14 39 1.006 0886 ~—-__08807 14 1.474 0676 += 04588 «15 40 994 0889 0890390 15 1444 0694 04804 16 41 983. 0801 ~— 09069 16 1416 0710 05014 17 42 971 0893-09196 17 1.389 0725 05219 18 43 960 0895 09321 18 1.363 .0739 05419 19 44 948 += 10806 += 00445 19 1339 0752 05615 20 45 938 0897 09566 20 1.316 .0764 05807 21 46 927 0898 09686 21 1294 10776 05994 22 47-917, —— 0899-0905 22 1273 «078 «= 06177.—«-23 48 907 .0900 09920 23 1.253 ~=—«.0796~— «0635624 49 897 0900 ~—_:10034 24 1.233. «0805» 0653225 50.887 0900 ~—- 10147 25 1.214 0814 .06704 26 Experiments on the Paranormal Cognition of Drawings [103 Statement of Method for the Non-Mathematical Reader For many years experiments in telepathy have made use of draw- ings upon which the agent concentrates and which the distant percipient tries to reproduce. Many such experiments have seemed to be successful because of numerous instances in which the per- cipient’s drawing closely resembled the object or drawing used by the agent. But there are several grave difficulties in attempting to assess the value of such results. First, the agent may have drawn a common- place object, or an object frequently thought of. Secondly, though the object be rare, there may be “parallel mental habits” shared by agent and percipient, so that as one follows a certain line of thought the other follows a similar line. Third, even if it could be shown beyond reasonable doubt that such an experiment as a whole is a success, one can never say just how successful it is. As one attempts to get control of such elusive phenomena, using a variety of methods, he may properly enough say that all the scores have been high, but whether one of his methods is better than the rest cannot be determined without a rather precise estimate of the degree to which the percipient’s drawings actually resemble those of the agent. ; The present procedure is an attempt to meet all these objections and to propose a simple and repeatable method which will, by and large, bring out clearly the degree of success achieved when free drawings are used. The first step of the present method is to ascer- tain from a large amount of existing experimental material just how commonly each object is actually drawn by a large number of percipients in a large number of experiments. We may find that there are, in the whole material, 281 lions, 2 zebras and no ant- eaters. In any new experiment in which a drawing of an animal by the percipient coincides with the agent’s use of a drawing of an animal, we may proceed to an exact assessment of the value of a hit. If the agent and percipient both give zebra the percipient gets very much more “credit” than if agent and percipient both give lion. The Catalogue presented in this book is nothing more than a list of all the objects drawn by percipients, showing how frequently each one is drawn. Now if chance only is at work the percipient may put down zebra at any point in the course of the experiment whether zebra is being actually used by the agent or not. Similarly with lion and the rest. If chance alone is at work, the percipient should be no more likely to call zebra when there really is a zebra than to call zebra when the agent is drawing a lion, an ant-eater, a telephone or a ship. Catalogue-scoring is a matter of counting up all the hits when the appropriate original was actually used on that occasion, and comparing this with all the correspondences between percipient’s drawing and drawing made by the agent on some other occasion.

RN aaa eS errr eee 3 104] Proceedings of the American Society for Psychical Resea F In the long run, if telepathy is at work, the correspondence of draw ings which are supposed to correspond will significantly exceed th correspondence of a percipient’s drawing with the drawing made hy the agent on some other occasion. 4 The scoring of a mass of data consists, then, in giving credit hits and ascertaining whether they significantly exceed pseudo-hits, that is, hits made on inappropriate originals. To achieve this the most important thing to know is exactly how much credit to give fo; each true hit; for some hits, as we have seen, are worth much more than others because the objects dealt with are rare, and a fey really brilliant ones may be worth more than many slight ones. Next we ask ourselves whether the value of a hit should go y in direct proportion to the rarity of the object drawn. At first sight it might seem so; but let us look at it this way: Suppose I am asked to determine clairvoyantly whether each of three pennies is lying heads up or tails up in a safe. This involves exercise of the clairvoyant faculty in three instances. To do it clairvoyantly thre times means three times as many clairvoyant acts as if there were one. If, however, I were to state the value of three correct guesses in terms of chance coincidence, I should merely say that I have one chance in two of getting the first right; I have one chance in four of getting the first two right; I have one chance in eight of gettin all three right (in other words once in two to the third power). The amount of clairvoyance goes up in equal arithmetical steps, 1, 2, 3, 4, etc., while the same degree of success would be expressed by stating the chances geometrically, 1, 2, 4, 8, etc. So for any situation the amount of clairvoyance present goes up not in direct proportior to the rarity of the object, but in proportion to the power to whicl one must raise the original fraction, such as, in the case describe one-half, one-half squared, one-half cubed; %4, %4, %, etc. If t chance of getting something is one in ten, and I make two successes in two such attempts, the result is not ten times as good as one success; as a measure of my clairvoyance it is only twice as goo So if I make one shot when odds are one in ten, one shot when thi are one in a hundred, and one when they are one in a thousand, my respective scores are in the ratio 1, 2, and 3. if Now this is all there is to the matter of using logarithms through- » out the paper. Logarithms are the simple standard way in which one specifies the power to which a number must be raised to give another number. In the formulae one will notice that the Fisher scores are always stated in terms of logarithms. i 9 We are now ready for the formula (cf. page 10). PY The number of hits is written h. We know what the chance of making a hit is, because we have the Catalogue at hand. The chance of making a hit is written p. We know how many attempts were made, having this degree of likelihood of being successful. This number is called m. In m attempts, we should pile up n times p successes, or np. We now want to find the excess of this beyond — Experiments on the Paranormal Cognition of Drawings [105 the number to be expected by chance, so subtracting we write h — np. We now put in the logarithmic expression which will measure the amount of credit which each such hit has earned. A moment's reflection will show that the smaller the chance (p) of getting a hit the greater the credit should be, so that to assign credit, p must be in the denominator of a fraction; and this will explain why instead of taking the logarithm of p we take the logarithm of one divided by p. The smaller the p, the greater the iractione will be. Thus when p is extremely small the logarithm of one over p will be ex- tremely large. : To restate again in words rather than mathematical symbols, we take the excess of hits over chance expectation, and assign to each hit a credit corresponding to the rareness of the object indicated; but instead of going up directly we go up as the powers of a given number. (What number we take is immaterial; it is customary in mathematics to write logarithms to a base of ten, saying for example that the log of 1000 is three, which means nothing more than that 1000 is 10 x 10 x 10, or 10°.) ’ One more logical step is needed. We want to know how this “credit” which we have assigned compares with the “credit” which we should have assigned to other items if they had been correct. It will be remembered that every percipient’s drawing must be com- pared with every drawing used by the agent. In other words, as one succeeds in aiming at a given object, one’s drawing must not succeed too well on all the other objects. To take a very extreme case as an example, it does me little good to specify lion as item No. 2 when that actually is item No. 2, if in drawing cards from a hat the agent should have drawn lion ten times. I want to compare my drawing not only with the intended original, but also with all the other items in the agent’s list. This computation of corre- spondences which are not credited as hits appears in the expression “variance.” The formula (cf, page 10) again makes use of m and 9, with both of which we are familiar, and with q which is simply / minus p. What it does is to take account of the total number of shots (n) made whether right or wrong, then multiplying both by p to indicate the chance of guessing it right and g to indicate the chance of getting it wrong, obtaining npq; but since we have thus put the product of two chance events into the formula, each of which is expressed by a fraction, we must multiply the logarithm by itself, which gives logarithm squared, log?. The = is the same as before. What we have done is to add up, throughout the whole experi- ment, all the “credits,” each being in proportion to the logarithm of its rarity, and then to compare this with the credits which would have been earned if they had been scattered randomly on a chance basis throughout the items used.

106] Proceedings of the American Society for Psychical Researe If the credits exceed a certain arbitrary standard (usually quantity which would be achieved once in 150 times) they are saij to be “significant.” Garpner Murpiy Postscript the above paper, I should like to add a few words based on in er. vening reflection and experience. ; I have no doubt at all that the drawings-technique as a whole, using any statistically valid method of assessment, is truly repeatal in the sense that anyone who cares to do what I have done will ob substantially the same results; though he may not, of course, if elects to do something different. That is to say, I should be on ground in retorting to any skeptic “If you don’t believe me, go and try for yourself”; and if he feels that collecting a few thousa drawings from a few hundred percipients is too much trouble, that is his misfortune rather than my fault. The method of assessment by Catalogue and Fisher Scores enormously more convenient and flexible than any system of “blind ' judging, depending on identifying “resemblances” ; and the Catalo here presented should prove at least a very useful yard-stick which the small-scale or casual experimenter can rapidly assess approximate value of his results, provided his sample of percipient is reasonably comparable with that of the Catalogue. It is necessar: however, that the main conditions of the experiments should be sub stantially the same as those of my own, notably as regards the us of fresh percipients making very approximately ten shots each at — sets of ten originals. ; This is a somewhat severe restriction, because it excludes, or at least discourages, the easiest type of work; namely, that in which a small group of percipients takes part in a series of experiments im each of which all or most of them perform. The reason is that, as experience shows, most people tend to reject any impressions they” may have of originals which have been used in earlier experiments even though they may have been explicitly told that these may be — used again in later experiments, as is necessary to maintain the principle of random selection. Consequently, if an original is used a second or third time, there will be a tendency for some percipients not to draw it, who might otherwise have done so, which handicaps — the experiment; whereas, if any or all of the percipients were to react in the opposite sense, and draw an object because it had already been used, they would not give themselves any advantage. I think Experiments on the Paranormal Cognition of Drawings [107 this can be remedied in principle by not telling any of the percipients what any of the originals were until after the whole series of experi- ments is finished; but not all percipients submit to this, while, even if they do, they will not be, in the second and subsequent experiments, in the same state of “freshness” as the percipients of the Catalogue, very few indeed of whom took part in more than one experiment, and then only after considerable intervals of time. The fundamental value of the Catalogue, however, lies in the fact that it is henceforward impossible for anyone who performs any kind of experiment with drawings (not diagrams), or for any critic of such an experiment, to say that he has “no idea” of how often the objects concerned would be likely to be drawn under chance condi- tions. The Catalogue will always enable him to form a pretty good estimate of the relevant frequency. Moreover, even as it stands, the Catalogue should enable any interested reader to carry out (as I hope some will) investigations of a by no means unimportant character, notably into what attributes of originals (strictly of objects depicted therein) are most favorable to the occurrence of the phenomenon. For example, he may enquire whether originals depicting a household utensil, or beginning with the letter C, or judged to be of an emotion-prompting nature, or com- monly met with in daily life, yield significantly higher scores than others not possessed of these attributes. It is the possibility of doing this kind of thing, which arises from the great variety and richness of drawings, that gives the technique, despite its initial cumbrousness, such great advantages as an instrument of research over methods using cards or the like. WHATELY CARINGTON