328 FLUIDITY AND PLASTICITY
small, the stirring ineffective and the end of the efflux tube is
exposed to the air. In making duplicate determinations the
liquid flows out into the air and generally cools off, so the bath is
raised to somewhat above the desired temperature in order to
bring the temperature back again to the large mass of oil in the
container. If the run is started when the temperature comes to
the proper point, it is almost impossible to prevent it going up
during the run.
Another important source of error arises from the very extra-
ordinary kinetic energy corrections encountered. The Engler
instrument, for example, is normally calibrated with water at
20°C and the kinetic energy correction amounts to over 90 per
cent of the total energy expended. The viscosity in this case has
but little part in determining the rate of flow, and we have already
seen that the coefficient (m) of the kinetic energy correction is
subject to some uncertainty.
Closely connected with the kinetic energy correction, are the
difficulties due to end effects and possible turbulence which are
aggravated in short, wide tubes.
j[? It is difficult to adequately clean this type of instrument or to
|l tell when it has been properly cleaned. The liquids readily
absorb dust, moisture and other impurities from the air and they
may thus undergo loss or chemical change. Meissner (1910) has
made a study of these sources of error. Effects of surface tension
at the end of the capillary, of the changing level of liquid in
the container, of slow drainage of oil down the side of the receiving
flask are found to be small sources of error. With the Saybolt
instrument, the flow is started by pulling out a stopper from the
hollow cylinder below the efflux tube. One must see that no
liquid accumulates in the air space above the stopper.
| Instruments embodying the principles worked out by Coulomb
and Couette have been devised by Doolittle, Stormer, and Mac-
Michael. In the Stormer instrument a cylinder is rotated by the
!| force arising from a falling weight, suspended by a cord carried
over a pulley. The speed varies with the viscosity of the liquid
and the revolutions per minute are counted. A better plan is the
one adopted by MacMichael of using a constant speed, imparted
to an outside cup and measuring the angle of torque produced in
a disk supported in the liquid by means of a steel wire. The