kanolt: melting points of kefractory oxides 317 



were made. The value of this fixed point accepted by the Bureau 

 of Standards is 1755°. The platinum was melted in magnesium 

 aluminate tubes in an iridium furnace, and both heating and 

 cooling curves were plotted. The weighted mean of the five 

 determinations was 1755°, with an average deviation of 5°. 



The problem of avoiding smoke when the oxides are not heated 

 in a vacuum has received very little attention in previous work, 

 and is probably responsible for many low results It was found 

 that errors from the presence of smoke could be avoided by 

 introducing into the oxide to be melted a tube of suitable material 

 into which the pyrometer was s'ghted, and keeping this tube clear 

 by a gentle current of gas. The cooling effect of the gas was 

 found to be quite negligible. With this apparatus it is impossi- 

 ble to see the oxide while it is melting, but as all melting points 

 were determined by heating curves, this is not necessary. All 

 determinations were checked by the examination of samples that 

 had been heated to temperatures slightly below that obtained for 

 the melting point, and also of samples heated slightly above the 

 melting point. 



Magnesia was melted in graphite crucibles, for it forms no 

 carbide at high temperatures. It does react with carbon with the 

 formation of carbon monoxide and magnesium vapor, as shown 

 by Watts; but this introduces no contamination into the mag- 

 nesia. The determinations were made at atmospheric pressure 

 to avoid too rapid vaporization of the magnesia and smoke was 

 avoided by the introduction of a graphite tube thru which gas 

 was blown. 



Lime forms carbide rapidly in contact with graphite at high 

 temperatures; graphite is therefore out of the question as a cruci- 

 ble material. Lime was successfully melted in tungsten crucibles 

 with an inner tube of tungsten, thru which hydrogen was blown. 

 Three results in excellent agreement were obtained in this way, 

 but the tungsten usually melted as soon as the lime had melted. 

 This was probably the result in part of contamination of the 

 tungsten by carbon, but the tungsten a' so appeared to be con- 

 taminated by contact with the fused lime, perhaps with the for- 

 mation of metallic calcium. This led to the suspicion that the 



