189 
more reliable, because the results derived trom the whole series 
are necessarily impaired by those of the uneven groups аз, 
moreover, is evident from the deviations of the differences from 
their theoretical value. 
The same remarks are applicable to the fourth element, the 
daily range of the temperature: here the results considered 
by themselves might be regarded as spurious because the group 
of uneven formulae does not show any trace of a periodical 
variation, but the fact that it is well marked in the even 
group and that the argument again agrees with the correspond- 
ing arguments of the two preceding elements, affords 
almost undeniable evidence for the reality of the phenomenon. 
The outcome of this discussion of the results obtained may 
justify the following way of deducing the most probable value 
of R, which necessarily is somewhat arbitrary : 
Daily means, barometer R= 25.8070 weight = 1.00 
idem, even groups 25.8040 0.50 
idem, uneven groups 25.8205 0.25 
Daily range, barometer 25.8252 0.50 
idem, even groups 25.7999 0.15 
idem, uneven groups 25.2544 0.15 
Daily means, therm. even groups 25.8045 0.50 
Daily range, therm. even groups 25.8155 0.50 
The mean value thus obtained is: 
В = 25.8120 days. 
If we adopt this value as the most reliable, the epoch, i. e. 
the phase of the curve on January the first 1866 at noon may 
be calculated by adding the angular quantities: 
— 299 x 12.9 £ —4.5 а 
— 529 x 12,92 — 5.0 а 
— 269 x 12.95 —4.02 
lo the arguments corresponding with р = 1 of the three groups: 
C 
25.812 & 25,800 
— 
— 
du 
