NUCLEATION CONSTANTS OF CORONAS. 8l 



the subsidence constant 5' = 6 . 5 is taken as the mea value of the above 

 data. To compute s = cm 1/3 /(6w/7r) 1 / 3 , the reduced values are 5 = 0. i6w 1/3 . 

 In table 36 the exhaustion ^ = 0.771 is smaller and the temperature 

 27. The constants have the corresponding values shown at the head of 

 the table. 



52. Remarks concerning the tables, and conclusion. The first series in 

 table 34 contains data both for 5, 0.12 5=5' and s, and leads to a cu- 

 rious consequence. The computed chords of the coronas, 5 = a(7rw/6m) 1/3 , 

 is not proportional to s = 2r sin but to S = 2R tan 0, where 26 is the 

 angular diameter of the coronas. This implies a diffraction equation read- 

 ing tan 6=1 .22 Xjd. 



These results are shown in fig. 26, where 5 aw 1 / 3 is laid off as the 



abscissas and 0.12 5 oc tan and o. i25/\/i +^S 2 /4^ 2 oc sin 6, as or- 

 dinates. If we confine our attention to values within 5 = 14, where the 

 readings are more certain, and where there is less accentuated over- 

 lapping of coronas, the graph 0.12 5 oscillates between two straight 

 lines as the coronas change from the red to the green types. The slopes 

 of these lines are respectively as 1.08 = 73.2^/0 and 0.99 = 73.2^/0, 

 whence ^ = 0.000047 and ^ 2 = 0.000043 cm. These should be blue and 

 violet minima. 



Fig. 26 shows, moreover, that compared with the graph for 0.12 

 5 = 6o tan 6, the curve for sin 6 is in series i quite out of the question, 

 as already specified. Hence in the remaining series of observations in 

 tables 35 and 36, 0.12 5 was used in place of 5. The results for the 

 series 2, 3, 4, are also given in fig. 26, in the same way. Curiously 

 enough, series 2 and 3, which should be identical with i, fail to coincide 

 with it in the region of higher coronas. In these series the graph 5 oc sin 

 would more nearly express the results, though the agreement is far from 

 satisfactory. Series 4 again corroborates series i, needing the s' ex tan 6 

 graph for its nearest expression; but in this series there is a curious 

 horizontal part corresponding to observed coronas of the fixed type 

 in the middle region of green coronas (5 = 10 to 12), showing that the 

 periodicity has been exaggerated. 



It is exceedingly difficult to account for this difference of behavior. 

 One may suppose that the phosphorus nuclei, which are here solutional 

 water nuclei, are not quite of the same size. This may happen if the 

 air is unequally saturated, for instance. In such a case the coronas 

 would be largest when the air is most nearly homogeneous and the 

 nuclei gradient within narrow limits (series 2 and 3), whereas in less 

 favorable cases (series i and 4) smaller coronas would appear. As the 

 abscissas, s = a (nTC/dm} 1 / 3 , where n z =y z ~ l K and the ordinates s (ob- 

 served) are independent of each other, the equality of 5' and 5 will in a 

 measure check the work apart from the constant a which determines . 

 This is actually the case for the lower series of coronas below 5-10. 



