[TURNBULL] A SUBJECTIVE PHENOMENON OF VISION 25 



plates having been " fixed " for a varying time before readings were 

 made; for instance, with the Y plate it was difficult to obtain a good 

 after-image without long " fixation," whereas the after-image for the 

 K plate continued for a considerable time, after a comparatively short 

 fixation of the plate — hence the periodicity, in this experiment, is 

 slower for Y tlian for R. 



52. The results of the experiments which follow (made from Feb- 

 rr.ary to April, 1896) are best seen from the curves. Plates I-V. Sev- 

 eral new colours besides the ones at first used were employed in these 

 experiments. Of course, with all the glasses used the light was far 

 from homogeneous, and the light transmitted by the 0, Y and C B 

 plates, especiall}^, was decidedly mixed ; however, the periodicity corres- 

 ponding to a certain sensation wasi what was sought and it may be that 

 the results with these glasses is not very misleading, but without doubt 

 homogeneous light of known wave-length would have been much more 

 desirable. 



53. In the following sets the Lambert apparatus placed in a 

 darkened room was used.. The transmitted light alone was used for 

 simple colours, the reflected light being shut off by ,a black screen placed 

 over the hole nearest the observer. 



54. Eests, usually of 5 or 10 minutes, were taken between succes- 

 sive sets. The chronograph (paragraphs 19 and 20) was used in all 

 the experiments that follow in place of the stop-watch (except where 

 otherwise stated). 



55. The abscissae (on all plates except Y) represent seconds from 

 the time of "fixing" the coloured plates (or other "backgrounds") 

 and the ordinates represent periodicities obtained by pressing the key 

 at the end of every 5 oscillations (and dividing the times thus recorded 

 on the paper strip by 5) from the time of beginning the counting. 

 Since the periodicities could not be obtained simultaneously, but had 

 to be deduced from 5 oscillations, the values for periodicities had to be 

 placed half way l^etween the times of starting and finishing the counting. 

 The curves were placed on different sections of the same plate to avoid 

 confusion of overlapping parts. 



56. Plate I represents the curves obtained in this manner in 

 single sets of readings, whereas Plates II-IY (inclusive) exhibit curves 

 platted from average results (the mîmber of sets making up the average 

 being usually ten). 



57. The upper curve (1) or Plate I represents the periodicities 

 for a run taken in the middle of the night (3.50 a.m., April, '96), 

 after awakening and before opening the eyes,, in a darkened room. It 

 will be noted that in this case the curve is smoother a.nd the oscilla- 



