NUCLEATION CONSTANTS OF CORONAS. 63 



There is another way in which the estimate in question may be made. 

 Let the nucleations corresponding to the colors be taken and reduction 

 made for the different drops of pressure in question. This is merely 

 a corroboration of the method of computation. The coincidence is as 

 close as may be expected, as the methods of approach are widely differ- 

 ent and the nucleation varies as the cube of the inverse diameter. 



Wilson's views of the nature of the phenomena are quite different 

 and lead to enormous nucleation, even as compared with the improbable 

 n 2l . He says (loc. cit., p. 301): 



When all diffraction colors disappear and the fog appears white from all points of 

 view, as it does when [the expansion] v 2 /v l amounts to about i . 44, we can not be far 

 wrong in assuming that the diameter of the drops does not exceed one wave-length in 

 the brightest part of the spectrum, that is, about 5 X io~ 5 cm. That the absence of 

 color is not due to the inequality of the drops is evident from the fact that the colors 

 are at their brightest when v 2 / v i ' 1S only slightly less than i . 44 and from the perfect 

 regularity of the color changes up to this point. 



Taking the diameter of the drops as 5Xio~ 5 cm., we obtain for the volume of each 

 drop about 6 X io~ H c. cm., or its mass is 6 X io~ 14 gram. 



Now, we have seen that when the expansion is such as produces the sensitive tint 

 (when v 2 /v l = i .42), the quantity of water which separates out is about 7.6Xio~ 6 

 gram in each cubic centimeter. With greater expansions rather more must separate 

 out. We therefore obtain as an inferior limit the number of drops, when v 2 /v 1 = i .44, 

 7 . 6 X io 8 /6 X io~ 14 = io 8 per cubic centimeter. 



In my data the smallest green corona corresponds to a diameter of 

 particles of about d 4 = 0.0005 2 cm., the next to d 3 0.00040 cm., the 

 next to J 4 = 0.00023, the first (which I have not been able to produce 

 by any means whatever, however large the nuclei) should correspond 

 to d l = 0.00013 cm., and even this calls for particles nearly three times 

 as large as Wilson's estimate (0.00005 cm.). In a small tube but 2 cm. 

 in diameter, like Wilson's test-tube apparatus, it is improbable that the 

 d 2 green corona, which is about 27 in angular diameter, could look 

 otherwise than greenish white, whereas the filmy disk of the large 

 crimson coronas (the largest producible, ^=0.00016, with an angular 

 diameter of about 39) would be mistaken for colorless. I shall venture 

 to believe, therefore, that Wilson's large greenish-white coronas corre- 

 sponded to about o . 9 X io 6 rather than to io 8 nuclei per cubic centimeter, 

 and that the maximum nucleation would not exceed io 7 even if colors of 

 the unapproachable first order were produced. 



36. Longer intervals between observations. Conclusion. Finally, experi- 

 ments were made with longer intervals of time, 2 minutes and 3 minutes, 

 between the observations. The object in view was the avoidance of 

 distortion of the higher coronas due to the absence of homogeneous nucle- 

 ated wet air in the fog chamber. But the longer intervals did not improve 

 the coronas and the data were for this reason discarded. 



