gg A CONTINUOUS RECORD OF ATMOSPHERIC NUCLEATION. 



data for the stopcock alternating with data for the instantaneous valve. Un- 

 fortunately, the time intervals between observations were not quite equal, as 

 the necessary adjustments consumed varying amounts of time. 



These data are shown graphically on the chart, figure 14, p. 83, the abscissa 

 being merely distributive. If the mean data of each set be taken the results are 



for the stopcock: 5 = 6.57, 53=284, 

 for the inst. valve: 5 = 7.04, 53 = 349, 



or the ratios of s 3 are about .81. Since, however, in view of periodicity, the 

 g-b-p coronas show abnormally small values of 5, it is better to compare the 

 red coronas only. In this case the mean values are 



for the stopcock: 5 = 7.0, 55 = 343, 



for the inst. valve: 5 = 7.2, ^ = 373. 



or the ratios of 5 3 are .92. Hence the two valves differ as to their data by less 

 than 10 per cent. If this value be ascribed by the value of m assumed it would 

 not suffice for the values obtained by photography. 



In table 18 the ratios of s 3 for the large crimson coronas under like occur- 

 rence were on the average .42, widely different from the present, showing that the 

 discrepancy in table 18 enters with the very small nuclei of the very large 

 coronas. 



One may note that the nucleation of the large room is gradually increasing, 

 due to a single small gas burner (source of light) , although the air was all pumped 

 into the condensation chamber from the floor of the room. 



40. Conclusion. As none of the explanations given are satisfactory, it 

 seems well to restate the case in conclusion with additional remarks on a possible 

 explanation. No matter which group of sizes of nuclei are efficient in producing 

 coronas, the effect of successive identical partial exhaustions must be to reduce 

 the numbers by the same relative amount. If a given corona is obtained for 

 the slower and the faster exhaustion for the same pressure difference, etc., it 

 may be assumed for argument that the same number of efficient nuclei must 

 occur in both cases, though they may not be the same nuclei as to size. It 

 follows that the same corona should occur in the two cases in each of the suc- 

 cessive exhaustions. This is only true below the middle g-b-p corona, i. e., for 

 relatively fewer nuclei (nucleation below about 100,000 per cub. cm.). In 

 this region of distribution, both valves eventually remove all nuclei, and they 

 remove them at the same rate. 



For nucleations above 100,000, however, proportionately more nuclei are 

 apparently removed by the slower exhaustion, cast, par., than by the faster 

 exhaustion, so far as coronal evidence is in question, precisely as if the faster 

 exhaustion were itself productive of nuclei (as, for instance, by breaking up 

 coarser into finer aggregates). One may note, moreover, that in this stage of 



