THE VJSCOMETER 69

means of Canada balsam which is warmed slightly until it fills
the end of the capillary.

If the capillary is elliptical, R4 in Eq. (20) must, according
to Riicker (cf. Thorpe and Rodger (1893)), be given the value

2B3C3

p2 , ^2 where B and C are the major and minor axes of the ellip-

tical cross-section. If the capillary is the frustrum of a circular
cone, Knibbs has shown that E4 must be replaced by

Ri2 + flifla + £22
where R and R2 are the radii of the two ends. If the capillary
is at the same time elliptical, S4 becomes
3fl33fi43 (1 - 62)3
E32 + R3Ri + #42 ' 1 + e2
where jR3 and R± represent the arithmetical means of the major
B — C
and minor radii at their respective ends, and e = w , n where
B and C represent the mean semi-axes. Knibbs has also con-
sidered the corrections necessary for other peculiarities in the
bore of the tube which need not be considered here.
There is no special advantage in using a variety of viscometers
for liquids of not very different fluidity. For liquids below the
boiling-point the fluidity never exceeds about 500. Assuming
this value as the maximum the lengths necessary for a capillary
of a given radius have been calculated by means of Eq. (5) and
plotted curve A in Pig. 24. It is not always possible to obtain
a capillary of an exactly specified radius, but with one having an
approximately satisfactory radius, the necessary length can be
read off from- the curve. For gases the maximum fluidity must
be taken as 10,000. If only very viscous liquids are to be meas-
ured the maximum may be taken as less than 500, curve B or C.
(cf. also Appendix A, Table IX.)
Construction and Calibration of Apparatus. — A point of great
importance in the construction of the viscometer is to have the
volume V (1) as nearly equal to that of V as possible, (2)
similar to it in shape, and (3) at the same height from the hori-
zontal capillary. This construction greatly facilitates the esti-
mation of the correction for hydrostatic pressure, within the