14 VAPOR NUCLEI AND IONS. 



proceed chronologically, taking the earliest experiments first. These 

 may be given with sufficient detail in a chart like figure 8, in which 

 the lower graph (I) shows the data obtained in the earlier memoirs 

 with a wooden fog chamber. The observed pressure differences (super- 

 saturations) are laid off horizontally, the apertures of the coronas 

 (5 = ^/30, nearly, where < is the angular radius to the outside of the first 

 ring, when the eye and the source of light are at distances 85 and 250 

 cm. on opposite sides of the fog chamber) and the nucleations vertically. 

 No particular attention was given to the occurrence of rain, and the 

 measurements cease at the lower end with the measurably visible 

 coronas. It is seen that the curve very soon reaches an asymptote, 

 showing hopelessly poor efficiency. As the work was carefully done, 

 the only explanation would seem to be that the speed of filtration was 

 insufficiently slow and the exhaustion insufficiently rapid (small vacuum 

 chamber). The same remarks apply to the graph (figure 8, II), ob- 

 tained in a decreasing march of pressure difference, at the beginning 

 of the work with the glass fog chamber (fig. 1); but the asymptote is 

 higher. The fog limit in this case is lower and attributable to nuclei 

 which have entered unobserved, as it is much below the values for 

 more carefully filtered air. The first observation, having been made 

 with the apparatus at rest for 12 or more hours, shows a very high but 

 probably more nearly correct result. 



Figure 8 also contains the corresponding nucleations, N, so far as 

 they can be computed, supposing that the nuclei are removed faster 

 than they can be restored. For high pressure differences the data are 

 necessarily estimates. Naturally the differences between the data 

 obtained after long waiting (12 hours or more), and those obtained in 

 succession after intervening intervals of a few minutes, are enormously 

 accentuated. 



It follows from these results that if normal data were to be reached 

 it would in the first place be necessary to perfect the method of filtration. 

 This was done by providing the filter with a very fine screw stopcock, 

 by which the rate of flow through the filter could be diminished very 

 gradually to a vanishing value. 



The next group of results, obtained on May 21, 23, and 30, and on 

 June 3, were found under conditions of very slow filtration. They are 

 sufficiently reproduced in the graphs (figs. 9, 10, 1 i),in which the abscissas 

 are the pressure differences and the ordinates the coronal diameters (s) 

 and the nucleations (n and N), respectively, as specified. The graphs for 

 5 are perhaps most suitable for discussion, although 5 exhibits the small 

 nucleations with very great advantage. The data for May 21 are in 

 the main much above those in fig. 8, but they are still far too low and 



