C. Barns — Apertures of Coronas. 339 



5. Cause of periodicity. — If one inquires into the cause of 

 the periodic discrepancies, it appears that the crimson coronas 

 are too small or else the green coronas too large, for the data 

 computed from exhaustions cannot be periodic. The former 

 being white-centered with a diffuse red margin, it is impossible 

 to mistake the outside edge of the first ring for the inside edge. 

 The blue-green coronas, however, show a uniformly colored 

 disc, and here the first ring may be of the same color as the 

 disc and the corona would then be measured to the outside 

 margin of the first ring. From this point of view only the 

 crimson coronas are adapted for measurement, and both curves 

 4 and 5 would then give d = 1*3 d' and ?i / = 2 , 2 n. Since the 

 curves actually give evidence of diminishing aperture while 

 the droplets certainly decrease in size, this explanation is plausi- 

 ble, though it does not agree well with the evidence from 

 normal coronas. The red and crimson coronas are the only 

 ones which retain the white center, and the phenomenon may 

 in so far be regarded as similar to the case of normal coronas. 



6. Effect of temperature. — The explanation of the discrep- 

 ancy between d and d' (computed from exhaustions and meas- 

 ured from apertures, respectively) reduces in the most favorable 

 case to d = 1*3 d', and for this two explanations must be exam- 

 ined. Supposing that one does not inadvertently measure into 

 a ring, the value of m which enters into the computation of d 

 is very variable with temperature. For 8p = 17 cm , for instance, 



at 10°, m = -42X 10~ 6 grams per cub. cm., 



•20°, -76X10" 6 " " " 



30°, 1*28 X10- 6 " " " 



Since d varies as m -1/3 , for the same nucleation,,the values 

 of d at 10°, 20°, 30°, will be in the ratio of -75, -96, 1-09, 

 respectively, and the coronas will be in the same degree 

 smaller. Between 20° and 30° this amounts to 2 per cent of 

 the value at 20°. Hence to bring the values of d computed 

 from successive exhaustion into coincidence with the data 

 computed from apertures, would require a temperature excess 

 of over 15°, which is out of the question. 



7. Pressure decrement. — As none of the explanations are 

 satisfactory, light of a different character maybe thrown upon 

 the discrepancy by computing the density ratio, y', of the gas 

 after and before exhaustion, corresponding to the observed 

 values, s'. Since if n is the nucleation, z the order of the 

 corona in a geometric series, b the coefficient of time loss, t the 

 time interval between exhaustions, 



log n = z (1 -f bt) log y 



the equation corresponding to a different exhaustion ratio 

 would be 



