270 Scientific Proceedings^ Royal Dublin Societij. 



of the coil remained sensibly constant. The power supplied ranged from 0-04 

 Avatt to 25 watts. 



Two sets of tests were carried out with the heating and cooling vessels filled 

 with water, but the connecting tvibe, or tubes, corked and empty. First the 

 initial rate of rise of temperature was noted when a known power was turned 

 on, so that the effective thermal capacity of the heating vessel and its contents 

 could be found. This allowed a small correction to be applied, when necessary, 

 for any slight changes of temperature during a convection test. Secondly, 

 by noting the final temperature attained by the heating vessel for a given 

 power, it was found that the loss of heat through the cork and the vacuum 

 vessel was proportional to the difference of temperature between the heating 

 vessel and the surrounding water-bath. A correction for this loss was thus 

 made. This was also small, except in cases of low power supply with a 

 narrow tube, when it became very serious. 



The stopper was then removed from the top of the tube, which was filled 

 with water, and the convection tests started. The usual procedure was to 

 turn on a certain current, start the cooling water flowing in the worms, 

 and leave the apparatus for several hours, tmtil a steady temperature had 

 been attained. A series of readings of the voltmeter, ammeter, galvanometer, 

 and thermometers over a period of 8 to 10 minutes was then made. From 

 these the rate of flow of heat up the tube for a given temperature difference 

 was found. In allowing for the effect of small variations in temperature the 

 effective thermal capacity of the heating vessel and contents was taken as 

 that already found with the convection tube empty plus half the volume of 

 the tube. 



The most satisfactory results were those obtained towards the middle of the 

 power range. For small powers the correction for heat loss becomes very 

 important and somewhat uncertain, as a very appreciable part of the heat 

 flowing up the lower part of the convection tube must escape laterally through 

 the cork jacket. On the other hand, near the top of the range the convection 

 currents sometimes became unsteady, causing the temperature difference 

 indicated by the couple to vary rapidly over a range of perhaps 20 per cent. 



Five sets of tests were carried out with single columns of various diameters 

 and lengths, and three sets with double columns, consisting of two parallel 

 glass tubes of as nearly as possible the same dimensions, with their ends at 

 the same level. In these cases circulation almost certainly occurred, warm 

 water rising in one tube and cool water descending in the other. The tubes 

 were separated by about 5 mm. of cork, which would fonn a fairly efficient 

 thermal insulation. 



The last pair of tubes tested being 28 cm. in diameter, necessitated the 

 use of larger heating and cooling vessels, a larger Dewar vessel, and a larger 

 floating vessel. In this case the heating coil was wound in several sections 

 in parallel, the joint resistance being 9 ohms, and the power range from 04 

 watt to 66 watts. 



As it was not possible to carry out all the tests with the same mean 

 temperature of the convecting column, some experiments were made with heated 

 water flowing through the worms in the water-bath and the cooling vessel. 

 In this way the average temperature of the convecting water was raised 

 without increasing the temperature gradient in it. The tests were in general 

 less satisfactory than those with cold circulating water, as the temperatures 

 were more variable. With a single column the rather surprising result was 

 obtained that the heat flow for a given gradient did not vary very much 

 for a range of mean temperature from 135° to 23-5° C. There was some 



