40 A CONTINUOUS RECORD OF ATMOSPHERIC NUCLEATION. 



What is ftirther menacing is the distortion produced by spherical and 

 cylindrical vessels, the surfaces of which are rarely quite concentric. In my 

 work with globes, I assumed that if the annuli showed no distortion and were 

 small in aperture as compared with the aperture of the globe, distortion could 

 be neglected. It is questionable, however, if this observation is vouched for, 

 since the apertures of coronas are peculiarly sensitive to refraction, particularly 

 when the distances of eye and source from the receiver are purposely chosen 

 large. 



Again, the quantity m is dependent on temperature. It is necessary 

 therefore to refer coronas to a standard teraperature as well as to a given degree 

 of supersaturation, and the correction is important if the coronas are to be used 

 in estimating the number of particles. 



Finally, the ratio of densities before and after exhaustion is a seriously 

 difficult datum to determine, for it depends on the degree to which adiabatic 

 conditions have been attained. It is here that the work is liable to be dis- 

 crepant. Hence a determination of apertures has an ulterior value, for it is not 

 improbable that the two series of results will muinally interpret each other. The 

 ]3resent chapter bears out this surmise, though it is merely to be regarded as a 

 rough test of my earlier results (/. c). An independent survey is made in 

 Chapter VI with plate-glass vessels. 



2. Apparatus and preliminary results. — The following charts contain a 

 preliminary survey of the sequence of coronas, their apertures, and the number 

 of particles of specified diameter encotmtered. The data for diameter and 

 number, d and n, are taken from m}- work on stxccessive exhaustion (/. c), where 

 the experiments are largely non-optical, and they are compared with the coiTe- 

 sponding data d' and n' which follow from measurements of aperture. The eye 

 and source of light are distant i and 3 meters, respectively, from the condensa- 

 tion chamber betweeii them. This was here a long c}'lindrical vessel of as clear 

 glass as possible, 50 cm. long and 13 cm. in diameter. The observations were 

 made parallel to the axis, absence of distortion being assumed for the axial plane, 

 an assumption which was justified by trial comparisons with plate-glass ap- 

 paratus, though the latter was not quite large enough for the complete survey. 

 The method of work was otherwise the same as that described in the earlier 

 papers. 



The results of the work may be given withotit tables in the accompanying 

 charts, figures 1,2,4, 5) 8, 9, in which the old results for d and n (computed from 

 successive exhaustions) are laid off horizontally, the new results d' and n' , 

 computed (as stated) from the observed aperture, vertically. The discrepancy 

 of the two sets of data is enormous, and the curves all show sustained period- 

 icity. All measurements of aperture, s, are made to the inner edge of the 

 red ring, and show the diameter of the central disc. 



3. Diameter of cloud particle. — The variations of d and d' are on the average 

 dd='i.4Sd', from curve 4, and 6d=i.6Sd' from curve 5. In other words, the 

 diameters obtained for coronas by computation from the conditions of sue- 



