338 G. H. KNIBBS. 
find 
os 
Table E. 
nm1:750 n=1°795 n = 1°850 
By Reynolds’ theory Hagen -250, Mair -205, Unwin +150 
By observation “344, 274, a) 2 
By calculation § 17 formula (35) 242 “222 199 
These results can hardly be regarded as conclusive, for though 
Hagen’s and Unwin’s Jend some colour to the view that the 
influence of temperature diminished as n increased, Mair’s are in 
direct opposition to that assumption. The law of diminution if 
true, must be left to future experiments to decide. The mean of 
the three results is 0°23, which perhaps ought to be used until the 
question has been decided. We shall however, return to this 
question later, vide § 17. 
14. Correction of k’, or U*, or of U for temperature.—From § 
12 it is evident that we may represent the actual observations by 
the expression 
=k fr (25) 
in which £’ is ks of the radius, and possibly of the roughness. 
Consequently 
ee OP iivciscses (26) 
The average of the temperatures in Table A. is about 14° C., and 
since the value of q is uncertain, we may calculate it by Reynolds’ 
formula, the application of which will at any rate partially correct 
the results furnished by experiment, to that mean temperature. 
And since the greatest difference is 10°, the correction will be 4 
small one. Logarithmically k, may be determined from /;, and 
similarly with regard to U* and U, thus :— 
log ky = log &, + q (log f, = log fy) ( 
log U;"= log U + a — log f) (28 
log U, = log U, + a (log A. - log f.) ( 
formule which are very readily applied. hk” has to be increased 
if expressed for a higher, diminished if for a lower temperature; 
assuming of course qg to be positive, 
