December 6, 1895.] 



SCIENCE. 



761 



suits, but would correspondingly lose in 

 compactness and portability. The appara- 

 ratus as illustrated has been used in the 

 physical laboratory at Columbia and has 

 proved itself very simple and usefid.* 

 Columbia College, November 18, 1895. 



3IEASUBE3IENTS OF THE ACCURACY OF 

 RECOLLECTION. 



We know that ordinary observation and 

 recollection are not altogether reliable. We 

 do not credit all the stories that we hear, 

 even though we may not doubt the good 

 faith of the narrators ; we see that conflict- 

 ing evidence is ofiered in courts of justice 

 when no perjury is intended; we regard 

 as partly mythical records supposed for 

 many centuries to describe historical events. 

 But we do not know how likely it is that 

 a piece of testimony is true, nor how the 

 degree of probability varies under different 

 conditions. If we could learn this by ex- 

 periment the result would be a contribution 

 to psychology, and would at the same time 

 have certain important practical applica- 

 tions. 



*The above drawing of the local variometer was 

 made by a method which may prove useful to others, 

 and which, so far as I know, is new. It is often de- 

 sirable to get a perspective line drawing of a rather 

 complicated piece of apparatus without employing a 

 skilled artist, or consuming too much time. To pro- 

 duce the above cut a small photographic negative of 

 the variometer was taken, from which a contact posi- 

 tive was made. The positive was placed in a projec- 

 tion lantern and thrown upon a screen consisting of a 

 piece of drawing paper upon a board, the size of the 

 image being two or three times as large as the cut. 

 The outlines were then traced in with pencil, and 

 one can also shade directly where desired. This 

 sketch was drawn over with India ink and made ready 

 for the photolithographer. The advantage of this 

 method over drawing ujjon a silver print, which is after- 

 ward ' dismissed ' is that the drawing and tracing is 

 done upon a larger picture than the final cut, and hence 

 a coarser style may be employed and yet the desired 

 finenesB attained in the final reduced cut. Of course 

 a projection lantern of some sort is desirable, but a 

 very simple one will do. The conventional shading 

 in the variometer was put in after the tracing was 

 finished. W. Hallock. 



I have tried in various ways to secure a 

 quantitative determination of the reliability 

 of recollection and evidence, and will here 

 report on the answers to some questions 

 asked the junior class in psj'chology in 

 Columbia College in March, 1893. The 

 questions were answered in all or in part 

 by the fifty-six students present. 



Several simple questions were first asked 

 and the students allowed in each case one- 

 half minute to consider and write the 

 answer. They were also requested to 

 assign the confidence which they felt in the 

 correctness of their answer — a if quite cer- 

 tain, h if tolerably certain, c if doubtful, d 

 if the answer were a guess. 



The first question was ' what was the 

 weather a week ago to-day ?' The answers 

 were pretty equally distributed over all 

 kinds of weather which ai-e possible at the 

 beginning of March. Of the 56 answei-s, 

 16 maybe classed as 'clear,' 12 'rain,' 7 

 ' snow,' 9 ' stormy,' 6 cloudy and 6 partly 

 stormy and partly clear.* It seems that 

 an average man with a moderate time for 

 reflection cannot state much better what 

 the weather was a week ago than what it 

 will be a week hence. Yet this is a question 

 that might naturally be asked in a court of 

 jiistice. An unscrupulous attorney can dis- 

 credit the statements of a truthful witness 

 by cunningly selected questions. The jury, 

 or at least the judge, should know how far 

 errors in recollection are normal and how 

 they vary under difiei-ent conditions. 



When asked ' what was the weather two 

 weeks ago? ' 20 students answered ' clear,' 

 and 18 'stormy.' The confidence in this 

 case was slight, only two being sure that 

 their answers were correct and 8 having 

 some confidence, while the others were 

 doubtful or did not answer at all. 



We ought not, indeed, to conclude from 

 these conflicting answers that no inference as 



* On the day in question it snowed in the morning 

 and cleared in the late afternoon. 



