480 White — Melting Point Methods at High Temperatures. 



ever been detected. The tubes used in our furnaces are of 

 two kinds, first an outer protecting jacket 8 min internal, 10 mm 

 external diameter, and closed below, separating the element 

 from the charge. These tubes also make a convenient support 

 for small crucibles, as already described. When so used, they 

 can be many times inserted in and withdrawn from a hot 

 furnace without cracking, and their life is almost indefinite 

 when not subjected to this severe treatment. When glazed, 

 they are much less durable, cracking rather easily, and cannot 

 be used with tight-fitting platinum jackets, but have the 

 important advantage of shielding the thermoelements from the 

 iridium vapor of the furnace. The other tube is an unglazed 

 open capillary of 2 mm outside diameter, for insulating the wires 

 of the thermoelement from each other. Ordinarily, one wire 

 of the thermoelement is left bare. This is necessary in com- 

 paring elements, and more convenient in most other cases, 

 except with the small crucibles, as explained above (p. 477). 



Recently, capillary tubes of " quartz glass " have been put 

 on the market, which are cheaper than the porcelain and 

 impervious to most gases. Where a curved tube is needed, 

 they are indispensable, for they can be bent (in the oxyhydro- 

 gen flame) more easily than ordinary glass tubing. Above 

 about 950° they cease to be permanent, slowly devitrify (crys- 

 tallize), and become brittle, but can still be used where their 

 special advantages justify the trouble and expense of frequent 

 renewals when necessary. 



To diminish the difficulties arising from contamination of 

 the thermoelements :* (1) Commercial platinum, with its high 

 content of iridium, must be banished from the furnace 

 altogether. (2) The elements used for the final measurements 

 are exposed to the high temperatures for the shortest time 

 possible. (3) They are also compared frequently with stand- 

 ards. (4) Great care is taken in standardizing to use the same 

 type of furnace and the same depth of immersion (almost to a 

 millimeter) as for the temperature measurement. (5) The 

 elements are cut off from time to time to remove the worst 

 contaminated portion. Of prime importance in this connection 

 is the factf that the reading of a thermoelement depends 

 mainly on the parts of it which lie in the steeper temperature 

 gradients. The pains taken to preserve constant depth of 

 immersion are for the sake of keeping the same portions of 

 wire exposed to the same temperature gradient. In our fur- 

 naces a difference in depth of immersion of one centimeter 

 may alter the error of a contaminated thermoelement from 20 

 to 30 per cent. Where contamination has gone too far, it is 



* Constancy of Thermoelements, W. P. White. Phys. Eev., xxiii, 463, 

 1906. Preprinted with important additions, Phys. Zeitschr., viii, 332, 1907. 

 fPhys. Eev., xxvi, 535, 1908. 



