408 
MESSRS. T. E. THORPE AND J. W. RODGER ON THE RELATIONS 
direction normal to the movement in such a way that strata at unit distance apart 
have velocities which differ by unity. The dimensions of -q are therefore [ML“^ T“^], 
It seemed advisable to design an apparatus which would admit of -the determi¬ 
nation, in absolute measure, of this coefficient for different substances and for a 
temperature range from 0 ° up to the ordinary boiling-point of the particular iic^uid. 
In this way instead of finding, as has been the usual custom, relative times of flow 
through the same apparatus under the same external conditions of temperature and 
pressure, and which might or might not be taken, as will be shown later, as measures 
of a single physical magnitude of the substance, that is, of its viscosity, the physical 
magnitude itself could be measured and the various influences which have been found 
to affect its value could be allowed for. The physical constants thus obtained could 
then be treated from the point of view of the chemist and the comparison would 
thus he of the same kind as that employed in connection with other j)hysical 
magnitudes, such as densities or refractive indices. 
Modes of Measuring Viscosity. 
Although the transpiration method has been almost exclusively used in researches of 
this kind, there are other methods of obtaining the value of iq. One of the oldest methods 
is due to Coulomb (‘Mem. de ITnst. Nat,,’ vol. 3, p. 261, 1800). It consisted in 
suspending a disc or cylinder within a mass of liquid and setting the disc or cylinder 
oscillating. From the diminution in the amplitude of the oscillations the value of the 
coefficient of viscosity may be calculated. 
Another method depending on observations of the oscillation of a liquid in a 
U-shaped tube was first proposed by Lambert (Mem, de I’Acad. de Berlin, 1784). 
The Coulomb method was modified by Maxwell. Piotrowski, at Helmholtz’s 
suggestion, instead of oscillating a regular solid in the liquid, obtained values of -q by 
oscillating a hollow sphere filled with the liquid; and quite recently 0. E. Meyer 
has shown that by the use of a hollow cylinder instead of a sphere, the accuracy 
obtainable in the theoretical treatment of the observations is considerably increased. 
None of these methods was suited for obtaining values of 17 over wide temperature 
ranges ; moreover, the large volume of liquid required to carry them out precluded 
their use in our case, owing to the difficulty of obtaining such large quantities 
of liquid in a state of sufficient purity. The tube method was therefore alone 
available for our purpose. It is satisfactory to note that Mutzel has obtained with 
the hollow cylinder a value for -q at 20 ° which is identical with that deduced from the 
tube observations of Poiseuille. 
General Principle of the Method and Description of the Apparatus 
Employed in this Investigation. 
The principle of the method employed by us consists in observing the time 
required for a definite volume of liquid, under a definite pressure, to pass through 
