128 G. H. KNIBBS. 
0-01786 : (38) 
It 79-0290 2 1 O00) 2F 
1 +.0-03395 7 + 0-000235 72 
The expression of the result to any higher order of precision wield 
be misleading. The general exactitude of Poiseuille’s experiments 
leaves little to be desired : it is to be regretted however that the 
dimensions of the tubes were not obtained to a much higher degree : 
of accuracy, and that the character of the thermometry has not been 
more explicitly stated. That it was very good is manifest both 
from an examination of his menioir and from a comparison of the 
results. For from (37) and (38) we have for the relations ofa — 
variation df” in relative fluidity and a variation bor of the vino 
to a variation in temperature dr 
df’ = (0-03395 + 00047 +) dr (39) 
— dy = (0-000606 — 0:0000328 7) dz (39a) 
hence a difference of 0°.1 will account for the difference between 
the mean and computed values of the fluidity at 0°.5 and at 45°, 
Moreover 0°.1 will alter the last figure in the coéfficient of viscosity 
for 0° C. by its whole amount. To determine the viscosity t0 
1/1000 the measurement of the temperature at 0° C. would have 
to be to within 0°.03, and at 100° to much less than 0°.01, so that 
to reach a higher order of precision than Poiseuille’s results seem 
to indicate has been attained, will involve an attentive consider 
tion of each element in a determination, and a complete discussion 
of the attainable accuracy at every stage of the processes of 
measurement. 
21. Other measurements of viscosity. Jacobson, 1860: 
Jacobson’s measurements of efflux were made with glass and 
brass tubes, the pressure under which the flow took place being 
that of the water in the reservoir of supply. His data in addi 
to the dimensions of the tubes are the ‘ heads’ and velocities th 
latter being computed from the weight of the efflux in a giv 
time, Hagen’s values! of the density of water at different tempe™ 
tures being used and the expansion of the glass and brass being 
oe 
1 Abhandl. Axed: Berlin, 1855, Math. Abt. p. 1-28. 
