336 G. H. KNIBBS. 
will hold good, or will at least very approximately represent the 
facts; hence taking logarithms, in order to test the relation, we have 
log ¢: # loge + g¢ log / . s16...5. (24) 
the equation of a straight line, determined by plotting log fas 
abscissee and log c’ as ordinates. Fig. 7 (a) shews the result : and 
q, the tangent of inclination with the axis of abscisse, proves in 
this case to be 0:244. The experimental justification of the 
assumption is remarkably exact, as the figure shews, and as the 
following vaiues for log c also indicate :— 
No. loge. Diff. 
1 44283 + 11 
2 309 —' 15 
3 284 +10 Mean 4:4294 = log of 26880 = ¢. 
4 314 = 20 
5 282 + 12 
More recently Mair' has also investigated the influence of temper- 
ature—between 13-°9 and 71-1 C.—-with a brass tube } inch radius 
and 25 feet in length. His manometer was 1 foot from the tank of 
supply, and apparently he regards this as giving the total fall in 
pressure for the 24 feet between the manometer and efflux end.” 
Mair observes that the plots of the logarithms of the heads and 
velocities give n=1-795 for his pipe throughout, the lines being 
parallel for all temperatures. The measurements are however not 
sufficiently exact to allow much weight to this assertion, since for 
example, I find from plotting his results for 13-°9, 48-°9, and for 
544 C., the values respectively of 1-782, 1:790, and 1°72. 
Accepting however the coéflicients which he himself deduces for 
the several temperatures of his observations, we obtain 
Temp. C. 13-°9 21:1 26:7 32:2 37-8 43-3 48-9 54:4 71°1 
Values of log f 0-180 -260 °316 -367 -417 -460 -501 “541 *643 
Value of log c’ 3559 ‘580 “590 -602 -613 -629 -640 “648 “686 
1 As tothe Effect of ac &e. ae last. C.E., Lond., Vol. 
LXXXIV., pp. 424 — 435, 1 
2 In such capone “ is far preferable to employ two manometer’, 
each sufficiently remeeney from the end to be unaffected by the terminal 
eonditions. 
