NO. 5 SOLAU \AU1 Altll.l IN Mtlior, |.U\VI.i:. AI.DUU II I I 



spoiulinj;" logarithms for tlio iiulividnal ilays. Wo tlu'ii plollod (In- 

 rrsultinj;' logarithms as orthnalcs, ami the positions on thr solai' (hsk 

 rathus as ahscissa'. In this way wc ohlainrd lOi iMth day <d' ohsi-iva- 

 tioii scvi'ti plots, or pairs of plots, in which tlu- slant of the n-pii'sciita 

 live liiK's iudicatod tlu" dcpartmc fiom tin- slaiidaitl nn-aii I'oiiditiou ol 

 solar i-oiilrast, at seven dilViMi'iit wavi- K'ni^ths. In llu- i<)i,^ rrdne 

 tioiis wc comhined the pairs of eheek valnes hefon- plollin-;, hnt ki'pl 

 thcni separate in the it)i ( n-dnelions. 



l*"ij;uie () shows the ii-snils ol two days ol eompaiison whiih are 

 typical of the two dilVeient classes of results ohl lined on tin- \arions 

 days. ( ^n .Septemher jj. H)!,^, tlu' cnrves inn iMncralK npwaid. 

 indicating a li'ss contiast than the mean, wlnlc on ( )clol)c-r jo. i<)I,^. 

 tlu- lini's iiin j;ciiorally downward, indicalin}; a j^realer contrast llian 

 the mean. Some em-ves on i\ieh (l;iy, to he snri". run coimter to llic 

 general Iri'iid of llie day. Iml these diver.mMici's aie le.isonahh lo he 

 rej^arded as oftiMi dne to acc-idi-ntal error. A ch.nitje of almos|)herie 

 tr;msp;irencv ol i p«'r cent dnrini; l\\c sini;le minnte of lime .it which 

 till" ceiilr.il p.iii ol tiic sim's disk is I'rossiiiL; the slit of llu- speciro 

 scope wonld aiconnl foi the divert^enci'S sho\\i\ hy wavi- liMii^lhs 

 O.67OJ and i.iK)S/A on Si-ptetnber jj. Still more prohahly the dis 

 crepancy may he accounted foi" hy a temporary shift of less than a 

 millimeli-r in the /vro of the holomeliic cnivi' I'ilher ;il ihe hcjMiiiiinL;, 

 middle, or end (d the iims. 



We desired to ohtain lor I'.ich d.iv's ohservalion ;i sinide v.iinc 

 which wonld he topical ol t-.ich d.iv's depailnri' (d contrast from llic 

 general mean. While loj^aiithmic cnrves of departmi's are not .always 

 smooth, and while there is no physical re.ison on whiih to h.ise ;i 

 prediction of the lorm \vliit"h they oni^hl to .issnmc. still in i;eiKTal 

 the cnrves are tolerahly ri-presented hy straight lines. Wc have 

 regarded the inclination of the best represi-ntative line foi e.icli 

 logarithmic plot .is .1 l.iir indication (d" the dep.irtin'e of coiili'.ist for 

 that day and that wave li'n,i;lli. Thric .ire seven wave lengths 

 ohservi'd on e.ich day of (thservat ion, .iiid we mii;ht h.ive t.iKcn Ihe 

 simple mean \ahu' of the l.in^ents of the incliii.ilion of llu- lo,i\iritliinit- 

 lines as the index of the deparlnre for the j;iven day. I'nt it appcari-d 

 that the depart nres were <^ii'.iter, the smaller the wave lengths, so 

 niiuli so lli.il the simple iiie.in wonld give entirely too nuich wi'iidit 

 to the shoilest wave length of ohservation. Accordingly, a system of 

 weij^hts has been deti-rmined hy comparing the magnitndt'S of Ihe 

 (lepartmes at llu- dilVerent wave lengths. The nniuhers for i<)i.^ 

 ;md l<)i.| were ohi.iined independeiillv from the data for lliose two 



