MR. H. TOMLINSON ON THE COEFFICIENT OF VISCOSITY OF AIR. 
773 
cylinders were made of paper wrapped round a metal core a sufficient number of times 
to secure the requisite stiffness ; the different layers of paper were pasted together, 
and when the whole was dry the metal core was withdrawn ; the outside of each of 
the cylinders was also coated with French polish to prevent the absorption of moisture. 
The mean diameter of each of the cylinders was measured by calipers reading to ToVo^ 1 
of an inch, and estimated to Toooo'fh °f an inch. In obtaining the value of the mean 
diameter of each cylinder, twenty measurements were made, ten at equal intervals 
along the whole length, and ten at the same intervals, but in a direction at right 
angles to the first. The measurements showed a very fair uniformity of diameter 
throughout the whole length, the mean being 1 '0079 inches for one cylinder and 
l’OlOS inches for the other. In the calculations subsequently made it was assumed 
that the diameter of each cylinder was the mean of the two last given, i.e., was 
1'0093 inches or 2'5636 centims. The lengths of the two cylinders were also very 
nearly the same, being 60'90 centims. and 60'85 centims. respectively; accordingly 
each cylinder was assumed to have a length of 60'875 centims. The ends of the 
cylinders consisted of wooden disks, into the centre of which was let a small brass 
disk provided with a screw, which was a companion to the screws at the ends of the 
suspenders S, K, so that the cylinders could be screwed right up to the disks M, M 
(fig. 3). The object of having the disks M, M, was to eliminate the effect of the 
friction of the air about the ends of the cylinders, # for Professor Stokes’s mathe¬ 
matical investigations only apply strictly to cylinders of infinite length. 
After the preliminary precautions previously mentioned had been taken the 
logarithmic decrement was determined from a great number of vibrations with the 
cylinders on ; the cylinders were then each turned round their axes through a right 
angle, for the purpose of eliminating any error which might otherwise arise from the 
section of the cylinder being slightly elliptical instead of circular, and the logarithmic 
decrement was once more found. The cylinders were now unscrewed from the 
suspenders, and, the brass caps having been for the purpose removed from the hollow 
bar W, the two brass cylinders k, h, were adjusted in the manner before mentioned, 
so that the vibratiomperiod might remain very nearly unaltered ; the caps were then 
replaced. All the adjustments alluded to above were performed very carefully so as 
to avoid jarring the wire, for if this precaution be not taken the internal friction will 
be temporarily increased, and will not come back to its previous value until the wire 
has been vibrated for a considerable time. A period of more than an hour was now 
allowed to elapse, the wire during this time being kept more or less in a state 
of vibration, but not through a greater arc than that represented by 400 scale- 
divisions from rest to rest, when the logarithmic decrement was again determined. 
* It would have been well to have bad these disks much thicker. As it is, the disks would only 
imperfectly serve the purpose for which they were intended ; the effect about the ends of the cylinders 
was, however, completely eliminated in Experiment IV. It would appear, moreover, from the results 
that with the long cylinders here used the effect mentioned above is neglectable. 
