E. L. DeForest on Correcting, ete. 371 
deficient in the green element, and therefore comparatively unfit 
to produce a bright green by mixture. 
A disc was now painted with alternate blue and yellow 
Sectors, the same materials being used; when it was caused to 
revolve rapidly the resultant tint presented to the naked eye 
was slightly reddish grey. Next the same dise was made to 
Totate before the slit of the spectroscope: prismatic analysis 
showed that its spectrum was totally different from the spectrum 
furnished by the green paper: it approximated to the normal 
Spectrum, being in fact the sum of the spectra y and B in the 
volving dise viewed by the naked eye appeared slightly reddish, 
Instead’ of pure grey. : 
It will be seen, then, that the results of spectral analysis ac- 
Count for this singular phenomenon in all i 
It would be equally easy to account for the fact, that much 
More brilliant greens are produced by mechanically mixing 
those yellow and blue pigments, which are, one or both, origin- 
ally slightly greenish in tint. 
New York, March 10, 1866. 
Art. XLV.—A method of Estimating and Correcting the Error 
caused by the unequal length of the Calendar Months, in reducing 
Observations of Temperature; by E. L. Dz Forest, of Water- 
town, Conn. 
THE mean daily temperatures throughout the year at any 
given place may be represented by ordinates to the curve whose 
equation is ; 
=A,+A, sin (x+E,)-+-A, sin (27+ E,)-+-A, sin (32--E,)-+&ke. 
the length of the year being 360° measured on the axis of «. 
The values of the first twelve constants may be found approxi- 
mately from the observed monthly means. But when these 
Means are made use of as if the calendar months were all equal 
in length, an error will be introduced into the equation. 
he amount of this error has been considerabl y underrated by 
Prof. J. D. Everett, of King’s College, Nova Scotia, in an article 
which appeared in this Journal for January, 1863. Supposing 
apparently that because February is the month which differs 
most from the other monthis in length, it must therefore be the 
immediate cause of the greater portion of the error, Prof. Ky- 
erett interchanged several days between it and the adjacent 
months January and March, and found that for Edinburg a dif- 
