724 PROFESSOR TAIT ON THERMAL AND ELECTRIC CONDUCTIVITY. 
to the other, so that at 300° C. it is little more than four degrees in advance 
instead of the five it had at zero, 
But, after all, this change of error (for the altered instruments were used 
for the higher temperatures only) can be easily allowed for in correcting the 
readings for scale-errors ; and it is very small in comparison with other 
inevitable errors of the determination. To mention only one of these, a very 
slight inexactitude in the position of the hole bored for one. of the higher 
thermometers would involve a more serious error. And, in the mercury, or 
fusible metal, in each hole there is a most peculiar distribution of temperature, 
due to the fact that one side of the hole is very considerably hotter than the 
other. 
§ 8. I have already mentioned the very great difficulty encountered in ob- 
taining a properly uniform source of heat in the statical experiment. I tried 
various processes depending on boiling points, and all sorts of gas regulators, 
without success, until I got a very valuable suggestion from Dr Crum Brown. 
The principle is excessively simple, but in working it was found to be almost 
perfect. It necessitated none of the constant watching described by Fores. 
All that was required was a reading of the whole set of thermometers every 
hour or half-hour. 
The following extract from my note-book tells its tale sufficiently :— 
Gas lit at 6.25 A.M, 
12h. 25m. P.M. 1h,1l1lm. th. 51m. 2h, 58m. 
Temperature at hole nearest source, 299 301 3011 301 
In fact, during the last three hours of the experiment referred to, the tem- 
perature, though about 300° C., varied by only about one-tenth of a degree. This. 
was actually less than the change of temperature of the air of the room, Of 
course this, and a few others like it, are exceptional cases; but not possible, 
even as such, with any other arrangement I have tried. As a rule, a change of 
at most three degrees in the temperature shown by the thermometer nearest 
the source (and this change a very gradual one) was the utmost fluctuation 
during the last three hours in the great majority of the experiments. In the 
few cases in which there was a greater change, it was traced at once to the 
‘‘burning-down” of one or more barrels of the six-barrelled Bunsen I employed. 
In such cases, the experiment was at once stopped, and the record crossed 
out. 
Nothing more satisfactory could have been expected in a matter so very 
difficult as that of regulating the gas supply, when, as all know, in a town like 
Edinburgh, the pressure is sometimes varied arbitrarily by an amount almost 
equal to one-third or one-fourth of the whole; and where, especially towards. 

