THE STRUCTTKE OF THE NUCLEUS. 31 



iiif a giv^eii exliaustioii lias a real inoaiiiiig, ami the cloud particles are as a rule so 

 fine that their subsidence during the obsei'vation [)eii()ds may be neglected except 

 when the final and normal coronas are approached. The interpi'etation of these 

 coi'onal sequences is now possible with a degree of pi'obability, by using the scale 

 of colors of Newton's interferences, with which they are liable to agree in succes- 

 sion. An excellent scale of this kind is given by Quincke and (quoted in Kohl- 

 rausch's Leitfaden (lev fmldUelien Pliysik. The cori'espondence of the colors of 

 coronal annuli, if worked nut in this way, will not be very perfect; but there is an 

 obvious general agreement of sequences, if the central field only of the corona is 

 taken. 



The axial color or coloi' of the full flame stated in table 4 is seen to make \\\) 

 a similai' sequence ahead in phase and nearly complementary to the central patch 

 of the corona. This will be separately investigated below, § 19 etseq. 



If the normal loss of nuclei in table 4 could be excluded, the succession of 

 coi-onas would represent a geometi'ic progression as to the number of nuclei, each 

 term corresponding to a number y —f/p^ or else {f/f^Y'^ times less than the 

 preceding. The advantage of this system will presently be shown, together with 

 an attempt to I'emove the time loss, which is also probably georaeti'ic. Meanwhile, 

 the corona of the s'" order is due to N— y' particles, relatively s])eaking. 



12. Lomof vnclel in the lapse of time (without exhaustion). — To ascertain in 

 how far such an interpretation as is given in the last paragraph is admissible and 

 to correct it for other simultaneous losses, it is necessary to determine the decrease 

 of nuclei when the receiver is left for long intervals, as far as possible without 

 exhaustion or other manual interference. In the following experiments the time 

 between the inevital)le exhaustions (usually about 96 sec, above) is prolonged to 

 thirty minutes or even an houi'. The first column of table 5 denotes the number 

 of the exhaustion (with refilling of filtei-ed air), the second the time at which it was 

 made, the third the minutes elapsed since nucleation. The fourth column gives the 

 color sequences of tiie coronas obtained, and these as a i-ule ai'e easily recognized in 

 order, by comparison with table 4. The coronal number in the series or ordei' is 

 put in the last column. Knowing the number of the corona, the approximate or 

 uncorrected residual number of nuclei (;Vjf>o)= luay be taken from table 4. The 

 remarks show that fogs are often spontaneously produced without change of pres- 

 sure (open stop-cock communicating with the atmosphere through the cotton filter). 

 The bearing of this on the present results is chiefly in the form of an error induced 

 by the subsidence of the loaded nuclei. Its theoretical bearing beyond this is also 

 to be kept in mind. Table 5 contains three independent experiments, in which the 

 times between exhaustions are varied. 



The next step to be taken is in refei'ence to the losses of nuclei inherent in the 

 method by which the data wei'e obtained. For orientation the data of table 5 con- 

 taining the residual nuclei, may be constructed in their dependence on the lapse of 

 time. The fraph, figure 4, is of the exponential character expected, with this differ- 

 ence, however, that tlie number of nuclei in the three cases does not diminish with the 

 frequency of the exhaustion. Thus in case (1) the number is even less than in case 



