NUCLEATION CONSTANTS OF CORONAS. 51 



in the earlier memoir. It is necessary, moreover, to investigate some 

 other method of obtaining 5 for the very large coronas, such as is given 

 in Chapter IV. In the present memoir the discrepancy is accentuated 

 by the short periods of i minute between the observations. This is not 

 sufficient for the complete mixture of the inflowing air and the nucleated 

 air within the fog chamber. As a result there are apt to be color dis- 

 tortions and bands of color before the real corona appears, while the 

 latter is not quite sharp. It was thought that longer intervals of waiting 

 between the exhaustions would have introduced other discrepancies 

 or losses of nuclei. Experiments made under these conditions did not, 

 however, much improve the irregularities, as may be seen in section 36. 

 Furthermore, in the larger coronas it is difficult to determine the actual 

 limits of the diffused annuli by the present single-source method. The 

 same difficulty will appear in the next section. Finally the d and 5 

 values computed from equations 1 1 and 1 2 show 



For the lower coronas these 5 values agree with the observed data 

 quite within the errors of observation, remembering that the coronas 

 were not perfectly sharp. For the higher coronas they are probably 

 close to the truth, provided the green and blue coronas be measured 

 to the purple rings. Both d and 5 will be discussed below and another 

 reduction will be attempted. 



29. Data for low exhaustions. Inasmuch as the only correction 

 added was for subsidence, it is necessary to test in how far convection 

 losses of nuclei upon evacuation, losses on evaporation, and losses in 

 the lapse of time (decay) are relatively small. This may be done by 

 comparing the data for very low exhaustions with the data for relatively 

 high exhaustions. In the former case many exhaustions must be made 

 and a longer time will elapse between the first and last of the equal 

 intervals than in the second case, where there will be relatively few 

 exhaustions and a relatively small lapse of time. If the errors in question 

 are negligible, the same initial nucleation and the same diameter of fog 

 particles for the same coronas will be obtained. The subsidence constant 



5 appears as follows: 



5 = 6.9 5.8 5.4 4.8 4.1 3-3 2.7 2.0 i.o 

 5= 15.4 2.6 5-5 6.7 6.8 3-9 3-9 



^ = 6.7 6.2 6.0 5.3 4-5 4- 3-3 2 -5 *-7 

 5= 4.1 7.6 8.4 3-8 5-8 5-5 4-* 



The mean values are 5 1 = 6-8, 5 2 = 4-9- Hence 5 = 5.9 was taken. 

 Experiments showed [dp 2 ] for 5 seconds of opening of the exhaust cock 

 to be equivalent to ^ = 0.873. The computed diameter 



