42 PROFESSOR A. CRICHTON MITCHELL ON THE 



diameter of this tube at the wider end was the same as that of the fan, and at its 

 narrower end 6 inches. But this small fan had been so constructed that it could only 

 drive air into the tube. I found that when this was done the air driven through the 

 tube had no steady motion ; that, in fact, its path resembled a spiral. 



The large wooden box into which the 30-inch fan was fitted was necessary for two 

 reasons. First, in order to get the most work out of such a fan for a given rate of 

 revolution it is necessary that it be allowed (what is termed by ventilating engineers) 

 'free feed' and 'free discharge.' In other words, it must catch up air from a free 

 space, and must discharge it into a free space. The wooden box fulfilled these condi- 

 tions sufficiently for all practical purposes. Second, the box acted as a regulator of the 

 speed of the air current through the tube when this might tend to vary owing to an}^ 

 slight irregularity of the gas engine which drove the fan. 



The end of the tube farthest from the fan was surrounded for a distance of 1^ feet 

 by a water jacket, M, into which water from a cistern entered at Q, and was discharged 

 a,t P. The temperature of the water entering the jacket was, in all the experiments, 

 nearly that of the air passing through the tubes. In any case where the temperatures 

 of air and water differed by more than 1° Centigrade the results of the experiments were 

 not employed in the final deduction of results. 



The copper ball, after being heated, was so suspended by its hook from a loop of 

 copper wire fastened to the inner surface of the tube that its centre was in the axis of 

 the tube, and in the middle section, P, of the jacketed portion of the tube. 



The instrument employed to determine the speed of the current of air was an 

 aluminium fan anemometer, constructed by Richard Freres of Paris. It was placed in 

 the position N shown in fig. 5, and so rested on three guides that its centre was in the 

 axis of the tube. It was found that with a given speed of the fan, the speed recorded 

 when the ball was hanging in the tube was lower than when the ball was not in the 

 tube. This difference, which was due to the ball disturbing the steady motion 

 of the air through the tube, was greater at higher speeds. Allowance was made 

 for it by directly observing its amount for each speed at which an experiment was 

 made. 



The temperature of the copper ball was' ascertained by means of a thermo-electric 

 junction of iron and German silver, which passed through a pumice-stone plug placed in 

 the circular hole bored in the ball. The junction of the two wires forming the circuit 

 was as nearly as possible at the centre of the ball. A Thomson's reflecting galvanometer 

 was included in the circuit. The value of unit deflection on the scale was ascertained 

 after each experiment by noting the deflection produced by placing the junction in 

 steam issuing from boiling water, and also noting the temperature of the mercury pool, 

 which formed the cold junction of the circuit. 



During the process of cooling, readings of the deflections on the galvanometer scale 

 were taken from time to time. As a rule, unit deflection on the scale represented a differ- 

 ence in temperature of 2° Centigrade. The error in reading the scale did not likely 



