1880.] On Actinometrical Observations, made in India. 157 



P S + x i» 



would be tabulated as if corresponding to the mean of the beginning 

 and ending times of X p ; no other times need be entered in the table 

 of individual results, which may thus be exhibited in a compact and 

 simple form. 



7. Remembering that the actinometer, as at present constructed, is 

 a differential instrument, it appears desirable to regard the causes, to 

 which the measured changes are due, under separate heads, which may 

 be briefly and generally indicated by 



Instrumental, 

 Local, and 



Intrinsic or Residual. 



[These three " heads " may be more definitely expressed symboli- 

 cally. If Y stands for the measured value of change in solar radiation, 

 suppose 



v.=s+(L+r)+(i+f) 



where S stands for that part of Y due to true or unmodified solar 

 energy, a quantity which appears unattainable by itself, at least 

 for the present: (L + Q and (I + i) are errors by which S may be 

 vitiated, so as to become Y. Of these errors, let (I-H) stand for the 

 portion due to the instrument and its manipulation, such that I can 

 practically be controlled and eliminated, while i may imperceptibly 

 exist in combination with S : also let (L + 1) denote all the errors 

 appertaining to the locality, i.e., between the sun and the instrument, 

 during the minute of observation, such that L, arising from visible or 

 otherwise detectable causes, can be excluded, while I, being invisible 

 and beyond endeavours to evade, may also imperceptibly exist in com- 

 bination with S. Then I and L stand respectively for the errors above 

 indicated by " Instrumental " and "Local." Now, if by skill, care, and 

 vigilance, we succeed in making L=0 and 1=0, and obviously acti- 

 nometrical observations are worthless unless there be reasonable pros- 

 pect of securing these conditions, then there remains, as the residual of 

 Y, the quantity 



y l= (S + Z + i), 



but since by definition I and i are not visible, nor yet detectable, their 

 presence or absence, in a given result, if generally suspected, can 

 neither be positively affirmed nor denied, appreciably speaking, nor 

 yet estimated. For their presence I use, as already said, the designa- 

 tion Y x (i.e., the residual of Y) ; in their absence, when Z=0 and i=0, 

 the designation "intrinsic" seems suitable, for now Y : becomes =S. 

 Hence arose the double designation, adopted above, of " intrinsic or 

 residual," whereby provision is made for either assumption, without an 

 vol. xxxi. n 



