282 1'ETTEUSHON, OK WATER AKD I C E. 



The determination of the change of volume by the melting 

 of the ice in experiment II miscarried. I was obliged there- 

 fore to start from the volume of the ice (= 1.089038) at — 10" C 

 found in experiment I, in calculating the vohimes of the frozen 

 water in Table II. It is evident that this circumstance does 

 not alter the whole character or the outline of curve II in 

 plate 22 in the slightest degree, although its position (i. e. the 

 €xact valne of the sp. volume of the ice) may be either a few 

 millionths higher or lower than the numbers given in Table II. 



From the above tables and from the graphic representa- 

 tion in plates 21 & 22 it will be seen, that also the purest ice, 

 which can be tested by experiments, is liable to premature 

 contraction of volume before melting. It is impossible to 

 decide, if absolutely pure water would be entirely free from 

 this weakness or not, since we can not assume that water, 

 which has boiled for a quarter of an hour or more in a glass 

 vessel, is absolutely free from minimal quantities of foreign 

 substances as f. ex. sodium salts, silicia etc. For my own 

 part I am rather inclined to think, that absolutely pure water, 

 if it could be tested, would show an absolutely fixed melting 

 point, but I think, that this problem very much resembles 

 another question still undecided, viz. is absolutely imre ivater a 

 conduding or non-conducting siibstance for electricity? In fa c t 

 K o hl r au sch has found, that the purer the water is, the 

 greater is the resistance, which it offers to the electric current, 

 and that we may by chemical purification and repeated distil- 

 lations etc. approach a point, Avhere the resistance tends to be 

 insurmountable for ever}" electromotoric force. In the beha- 

 vior of solid bodies just before their melting points I think I 

 have found another, almost equally sensible, proof of their 

 chemical purity. The regular coefficient of expansion of pure 

 water in the solid state (see Table I) slowly increases from 

 0.000165 [the average value between — 17° C and — 10*^0] to 

 0.000171 or 0.000174 [between — 4° and — 3° C] but then gra- 

 dually begins to decrease and finally changes its sign between 

 — O^.is and — 0°.o3 C, where the ice begins to contract instead 

 of expanding its volume. This point of inversion is reached 

 by water II already between — 0°.3o and — 0^05 C and by III 

 between — 0°.35 and — 0°.25. In order to obtain a stricter 

 comparison we must refer to the preceding tables, as the gra- 

 phic representation on plate 22 does not give us a clear idea 

 of variations beyond O.ooooi of the unit of volume [1 cc. of the 

 water at 0°] 



