1138 



indeed below tlieir melting-point, but it was impossible to make 

 them into wii-es. Hence they could not be investigated conformably 

 to the method we made use of. 



The W, Mo, Fe, Cu, Au, Ag, Mg, Ccl and Zn we made use 

 of, may be looked upon as chemically pure; this was borne out by 

 analysis and measuring of the temperature-coefficient of the resistance 

 (e.g. with Au 39.8, with x\g 39.2). For carbon we used the well- 

 known carbon-filament lamp wire. Two kinds of platinum were 

 used: one being extremely pure, a second containing something of 

 the other Pt metals as impurity. The A'i-wire contained 99,4 "/o ^^^ 

 and besides traces of Co, Fe, Zn and .4/. 



With base metals slight oxidizing of the wire, e.g. with the leading-in 

 of the support into the glass cover can generally not be avoided. 

 For this reason hydrogen was admitted into the balloon, after joining 

 the lamp to the airpump, and the wire in this atmosphere was brought 

 to a slight incandescence and then it was evacuated once more in 

 the usual manner. With some very low-melting metals special lamp- 

 shapes were applied, in order to prevent a too gieat heating of the 

 metal to be tested during the fusing into the balloon. 



^ 4. Silver. 



Concerning this metal many observations have been made during 

 the first investigation. It struck us, that though the Canadabalsam 

 suflficiently arrests the direct change of colour in the deposit of 

 metal that is observed with access of air, yet it cannot eventually 

 prevent it. It niay be that it is not altogether indifferent itself, or 

 that it does not altogether arrest the diffusion of the air-gases, at 

 any rate it has some effect in the long run, and therefore it was 

 desirable, in order to determine the nature of the unaffected deposit 

 of metal to control the observation with a Canadabalsam-coating or 

 with some other protector. Our choice fell on the salt-films formed 

 by sublimation, concerning which, one of us ') had observed during 

 previous investigations, that it does not perceptibly effect a tungsten- 

 film that is already deposited. Having observed that the sublimated 

 salt forms a perfectly homogeneous, optically insoluble layer of glass, 

 we might expect it not to interfere by a structure of its own, with 

 the observations of the metal. Among the various salts calcium- 

 fluoride ') was found to be preferable, on account of its insolubility 

 in water. In order to fix the layer of salt, we wound, in lamp c 

 (fig. 1) across a second set of supporting-hooks W-wire with separate 



1) L. Hamburger Ghem. Weekblad 13 (1916) 535. 



^) L. Hamburger en D. Lely. Ned. Octrooi-aanvrage no. 6502. 



