851) 



According (o llii^ liigli (lo,i2,rec of [mrily of llie coiiipoiiciits, all 

 the binary mixtures, eni|)loyc(l in this rcsearcli, wei'o al)s()liilely 

 white. They were obtained by heating weighed ({uantilies of both 

 oxides in a finely divided state in platinum or nickel crucibles ; this 

 manipulation was done in small resistance-furnaces, at temperatures 

 ranging from 900^ to 1000° C, and every contact with a reducing 

 atmosphere was carefully avoided. After grinding down the masses 

 and sifting, they were treated in just the same way, etc., till the 

 whole preparation was shown to have a homogeneous composition. 

 Mixtures between ortho-, and meta-silicate were prepared from both 

 these compounds in quite the same way. The preparations were 

 analysed after Hillebkand's indications'), and always in duplo. 



Every admixture of formed nickel-oxide was carefully avoided ; 

 mixtures rich in lithiumoxide however, needed to be heated in platinum 

 crucibles, because they would dissolve otherwise tinely divided nickel, 

 which coloured the preparation wilh a beautiful \iolet hue. For 

 instance the orthosilicate could be obtained in this way, as a splendid, 

 intensively coloured, violet product. The study of mixtures, richer 

 in lithiumoxide, corresponding with the comj)osition of the ortho- 

 silicate was not possible in the usual way, both because of the vola- 

 tility of the oxide at the prevailing temperatures, and by the fact, 

 that the lithiumperoxide LijO^, generated at higher temperatures, will 

 very quickl}' attack the platinum and the wires of the thermoelements; 

 the platinum is superlicially coated with a dull olive-green or greenish- 

 fellow layer, and every accurate temperature-measurement thus gets 

 practically impossible. Experiments, made in hermetically closed 

 platinum-vessels however, ha\e given many good results, as will be 

 described further on, in § 12 of this paper. 



§ 3. The temperatures of equilibrium were determined in the 

 way always used in this laboratory. A WoLFF-potenliometer ij-esis- 

 tance : 83 Ohm) with three decades, and with a constant resistance of 

 the galvanometer-circuit (White — Diesseliiorst), was employed, in 

 connection with a sensitive AYRTON-MATHEH-movingcoilgalvanometer, 

 with high resistance, and short period, to compensate the momenta- 

 neons electromotive force of the thermoelement ; the galvanometer 

 was calibrated and adjusted in such a way, that one microvolt 

 corresponded with a millimeter on the scale. The observations, made 

 with our thermoelements I and 11 (platinum and platinumrhodium 

 (]07o)-alloy) could be compared immediately with the nitrogengas- 

 thermometer, by means of calibration with a standardelement, cali- 



VV. F. HiLLEBRAND. Analysis oi Silicate and Carbonate Rocks, 2nd. Edit. 1910. 



