p. LECOMTE DU NOUY 



745 



Langley gives another series of figures (2)^ by which he intends to express "the 

 proportionate results for seven points in the normal spectrum, whose wave lengths 

 correspond approximately with those of the ordinary color divisions, where unity 



1 



is the amount of energy (about 7ZZ^ erg) required to make us see light in the 



crimson of the spectrum near A." According to this definition, this scale corre- 

 sponds to the minimum visihile. 



Wave lengths. 



Luminosity (visual effect). 



0.40 



1.600 



0.47 



62.000 



0.53 



100.000 



0.58 



28.000 



0.60 



14.000 



0.65 



1.200 



0.75 



Expressed in negative powers of 10, in order to facilitate comparison, we have 

 (unity being 10~^ ergs, no indication being given concerning the time): 



The first figure (for 0.40m) agrees well with that given by Langley (2)^ in his 

 other tables. The second one (0.53m) does not agree at all, and the slight 

 difference in wave length cannot be regarded as the cause of the discrepancy. 

 The third one agrees within 25 per cent and the last one also, approximately. 

 We see no explanation for this discrepancy, which cannot be due to a misprint. 



Therefore, it was desirable to settle the question, since Langley's data are so 

 misleading that good authors have made errors simply in quoting them. Quite 

 recently, Joly (8) published a very interesting article on a quantum theory of 

 vision, and although he does not share Henri's opinion on the subject, quotes 

 one of his figures, 5 X 10"^^ ergs for the threshold of sensitivity for white light. 

 Now, we have tried in vain to find such a figure in two of the papers of Henri 

 and des Bancels, as the indications of the source are missing. As far as we 

 know, they did not make any measurements themselves, but simply quoted those 

 of Grijns and Noyons. They quote the figures given by Grijns and Noyons, 

 4.4 X IQ-" ergs. Even if we admit that only 10 per cent of the energy is radi- 

 ated under the form of light (9), we obtain 3.96 X 10-^^ and not 5 X IQ-^". 

 It is regrettable that Professor Joly did not give the bibliographic reference. 



Method. 



An integration method was used. In other words, a curve repre- 

 senting the intensities of the dispersed beam after its passage through 

 the prism was plotted in function of the wave lengths on coordinate 

 paper. It is clear that the area delimited by this curve and certain 



