48 PROCEEDINGS OF THE AMERICAN ACADEMY. 



Was a Stable Condition Reached 1 ? 



The question how long the Thomson effect apparatus would require in 

 order to reach a practically stable condition of temperature, after the heat- 

 ing at one end and the cooling at the other had begun, was one which pre- 

 sented itself and was considered long before the apparatus was set up. 

 With the assumption of no lateral flow of heat from the bars, calculation, 

 by a well-known use of Fourier's series, showed that two hours would be 

 ample time. But such a calculation could not be decisive for the actual 

 conditions of the experiment, where lateral flow did exist to a considera- 

 ble extent, and where the thermal condition of the board B (Figure 6) 

 might affect in some measure the thermo-electric currents passing through 

 the wires lying upon it. The usual practise in a day's work was to have 

 the thermal conditions at the ends of the apparatus established at least 

 two hours, and the electric current in the bars at least one hour, before 

 observations on the Thomson effect were begun ; but it was not taken 

 for granted that a stable condition was reached by this preliminary de- 

 lay. On the contrary, of the six cycles, or runs, made, as a rule, each 

 day (see p. 31), the first cycle and the first cycle after reversal of the 

 current through the bars, which latter cycle usually began about one 

 hour after the reversal, were at first regarded with much suspicion. 

 Accordingly, the mean result obtained by a rather summary calculation 

 from these suspected cycles in fourteen * days of work was compared with 

 the mean result obtained, in a similar way, from all the cycles of these 

 fourteen days. The difference of the two means was found to be less 

 than one third of one per cent of either mean, and therefore all cycles 

 were treated as entitled to equal confidence. 



Nevertheless, observations made with the express purpose of getting 

 data for an estimation of the lateral flow of heat from the apparatus 

 showed that perceptible progressive changes of gradient were going on 

 in the bars after several hours of fixed terminal condition. These 

 changes were probably not of a character to effect perceptibly the differ- 

 ential of gradients, with which we are mainly concerned. 



The Lateral Loss of Heat. 



This loss has already been referred to, and the general method by 

 which it was estimated has been given on pp. 34 and 35 ; but some details 



* There were fifteen such days of work ; but the data of the first day were not 

 sufficiently full to be used here. 



