MR. H. TOMLINSON ON THE COEFFICIENT OF VISCOSITY OF AIR. 
771 
box B, whilst the mean of the readings of T x and T : , was used for the temperature of 
the wire. The thermometer T 3 was divided to one-tenth of a degree Centigrade, and 
had been tested at Kew ; whilst the thermometers and To, which were graduated 
in degrees Centigrade, had been carefully compared by myself, degree by degree, 
with T 3 . 
The barometric pressure was registered by means of a delicate aneroid barometer, 
reading to y^tli of an inch, which has been in my possession for 15 years ; this 
instrument I had recently compared with a standard mercury barometer.* 
Before commencing the actual experiments on the viscosity of the air, it was found 
advisable to subject the wire to a preliminary training, in order not merely to 
diminish the internal friction of it, but also to make this last as constant as possible. 
In the first place, the wire was well annealed; this had the effect of reducing the 
internal friction of the hard-drawn metal to less than one-half of its previous 
amount.! In the next place, a load, equal to that of the cylinders or spheres to be 
used, having been suspended to VV, the wire was alternately heated to 100° C. and 
cooled again, this process being repeated for about a week, on each day of the week, 
until there was no further alteration of the internal friction of the wire when cool. 
This treatment still further reduced very considerably the damping of the vibrations 
due to the wire. The manner in which the heating’ was effected will be shown in a 
future paper, in which also will be recorded the results of experiments on the tem¬ 
porary effect of change of temperature on the torsional elasticity and internal friction 
of the metals used. When the wire had undergone this preliminary treatment, and 
all the arrangements were complete, the bar VV, with its appendant cylinders or 
spheres, as the case might be, was started by small impulses imparted by a worsted 
thread, until the arc of vibration, as reckoned from rest to rest, had reached about 
400 divisions of the scale (about 10°, since 41’227 divisions represented 1°). After 
the arc of vibration from rest to rest had subsided to about 200 scale-divisions, the 
vibrator was again started, and. this process was repeated until something like a 
thousand oscillations had been executed, j Finally the vibrator was re-started for 
the actual observations, through an arc of about 200 scale-divisions, and when 
about 50 oscillations had been executed after this last starting the readings were 
begun. Suppose that a lt b Y ; a. 2 , b. 2 ; a 3 , b 3 ; a 4 , ; a 5 , b-, and a 6 are eleven consecutive 
* In spite of tile long pei’iod which has elapsed since this instrument was first made for me by the 
late Mr. Becker, of Elliott Bros., the spring still shows a slight amount of permanent yielding, which 
during the last two years has altered the reading by '015 inch. 
t Either silver, platinum, or copper wires, well annealed, may he nsed with advantage. T should not 
recommend unannealed piano-steel wire as used by Maxwell ; the last metal possesses, it is true, great 
elasticity, but the internal friction of silver, platinum, or copper can, by annealing, be made considerably 
less than that of the unannealed steel. 
X The object of this treatment was to reduce the internal friction to its permanent condition, since 
long rest, or sometimes even a comparatively short rest, always raised sub-permanently the internal 
friction. 
0 F 2 
