( 254 ) 
also the melting-point line of Na 2 S0 4 . 10 H 2 0,BE and that of 
Na 2 S0 4 .7 H 2 0 , CD could be derived already before from the deter¬ 
minations of Gay-Lussac x ), Lobwkl, de Coppet, and Berkeley 2 )., 
Further Gay-Lussac and Berkeley have also made investigations 
on the melting-point line of the anhydride, and pursuing their deter¬ 
minations up to ±100°, they could already state that we meet here 
with the special case of a retrogressive melting-point line. 
The first observations above 100° were carried out by Tilden and 
Shenstone *) from 100° to 230°, and by Etard 4 ) from 150 to 320°. 
Their results have been joined in the following tables. 
TILDEN and SHENSTONE. 
Tilden and Shenstone’s table, which is sufficiently accurate to 
justify us in deriving conclusions from it, shows that the melting- 
point line possesses a vertical tangent between 100° and 140°. 
Only one of Etard’s determinations has been serviceable, viz. that 
at 320°, for this led us to suspect that the plaitpoint curve would 
meet the melting-point curve, which, as was mentioned in the be¬ 
ginning, appeared really to be the case. 
In the accurate determination of the melting-point line from 62 
to 365° (according to the method of small tubes) 6 ) we succeeded in 
drawing up the most interesting part of the Tsr-figure with perfect 
certainty, and in demonstrating that the system H s O — Na 2 SO* 
J ) Ann. de Chim, et phys. 11, 312 (1819). 
-) Trans. Roy. Soc. 203 A. 209 (1904), 
s ) Trans. Roy. Soc. 175, 23 (1884). 
4 ) G. R. 106, 206 (1888). 
» „ 113, 854 (1891). 
5 ) Above 250'’ quartz tubes were used. 
