( 414 ) 



The second term — the Jovian evection, whose coefficient given 

 here according to Radau agrees ahiiost exactly with Hill's value — 

 and the 4^^^ term give rise in the coefficients of sin (j and cos g to 

 terms of a period of about 18 years (the exact periods are 17.41 

 and 18.61 years respectively), while the periods of 2 rr -|- 3 [" — 5^ 

 and of 2 .T — ^J-^l" amount to 9.74 and 37.25 years respectively, 

 i. e. about half and double the length of the former. 



The combined influence of the theoretical terms 11 and IV therefore 

 must be represented approximately hx Newcomb's and my empirical 

 term, for which I found a period of 18.6 years, and the other terms 

 can have but little influence on its determination. In one respect, 

 however, the method followed in our computations will be erroneous: 

 we have wrongly assumed that h' and /' ha\e the same argument 

 as well as the same coefticient. 



To investigate in how far the dilïerent theoretical terms were 

 confirmed by the observations I \\u\q proceeded as follows. On the 

 one hand 1 have tried to find Avhether the formulae determined 

 originally for It and Jt, where the equality of argument and coefficient 

 was not yet assumed, point to the existence of the term IV. On the 

 other hand I have investigated whether the diflerences found before 

 between the observed and the computed h and /• [toint to the existence 

 of the terms I and III. For their influencö must be exhibited in 

 those differences. 



If we take 1876.0 as zero epoch and assume foi' the annual 

 variation of the argument the value finally foujid -j- J^ ••^•^' ^'^^ 

 first formulae derived on }). 379 of my first pa])er are. 



h= ^ 0".45 — 1".30 sin [317°.l + 19^35 {t — 1876. 0)J 

 A- = + 0".26 + 1".46 cos [299°.3 + 19°.35 {t — 187(3.0)] 



If we assume that the variable i)art of h must consist in: l"^' . a 

 term which agrees in argument and coefficient with the variable 

 part of h, and 2"*i. a term of the form IV, we find for the latter: 



Fundamental Catalogue this becomes -|-0".13. Newcomb found (Invest, p. 33) for 

 1862—73 as mean difference: Declination observed at Greenwich — Tabular Decli- 

 nation + 0".36. Constant corrections had already been applied to the observations 

 by Newcomb and, if accormting for them I now reduce the results to Newcomb's 

 f". C, I find Obs.— Comp. = — 0".U8. So the constant correction to be applied 

 to the tabular declinations is found to be small for the two periods. (Comp. also 

 below secfion 18 last part). 



If the observations were reduced with Newcomb's value of the moon's parallax, 

 the differences Obs.— Comp. would be about + 0".44 and + Ü".23 {la-sf part 

 added 1903 Dec.) 



