136 G. H. KNIBBS. 
mpPT R* 
64 L? 95 
which is quite satisfactory since the corrective term is small. 
Pen P (1 +ar+ Br?)? Pisisteacsenees (44) 
22. The value of »o.—The value of the viscosity of water a Orc. 
—viz. 79 = 0-01786 formula (38) corresponding to = 0-013103, 
and deduced by a rigorous reduction of Poiseuille’s efflux measure — 
ments—is among the smallest of those obtained from the exper: 
ments of different observers, see the second table of § 6. This is 
probably attributable to the fact that, of all tubes of equal 4 
sectional area, the right circular cylinder gives the greats) — 
discharge, whence it follows that where the radii have beet 
computed from the measurements of capacity on an erroneous 
assumption of that form, the departures therefrom must necessarily, 
as shewn, in § 11 to § 16, give values greater than the true me — 
Poiseuille’s tubes seem to have been the only ones whose sections — 
were measured directly, and at several points along the ax 
consequently it is to be expected that the value of », deduced 7 
from his experimental data by the application of corrections fot 4 
the departure from the circular cylindrical form, will be appros 7 
mately the minimum. Its smallness, as compared with the resull 
of other observers whose data do not include material for the 
corrections, in no way diminishes our confidence in it as 4 relia! 
value, and one really representing, in numerical form, a phys! 
property of water; whereas in regard to the larger values one very 
properly suspects that had such data been to hand, the appropriate, 
corrections, then possible, would have made the results more 
accordant, since they are always negative in sign. At the a 
time we do not prefer the smallest value because errors ¢ 
measurement, either of dimensions, pressures or times, are aS likely 
to be negative as positive. 
For purposes of comparison the following values of the viscos!t] 
at 0° C. have been recomputed from the original data using 
Boussinesq’s value for m in (5), the reduction to 0° being 
by multiplying by the relative fluidity f’—formula (37). 
temperatures quoted are those of the experimnts. 
