314 ; Scientific Intelligence. 
proved, there is always, even in extreme cases, a tendency to 
exclude such material, a tendency which is the more effectual in _ 
rity may, i c at 
; eiphines naan fvalal a good Cieneniicn of the theory we are 
discussing. Water and oil of vitriol may be mixed in any pro- 
portion and there is a perfect continuity between the various 
stages of dilution. When bowever on exposure to cold the defi- 
nite hydrate H,SO,.H,O crystallizes out this continuity is ab- 
tuptly broken, and it is reasonable to suppose that the molecules 
» this compound preéxisted in the liquid = only aggregated 
the of crystallization. These stals may be ob- 
deed tween quite wide limits of Sigton and must enclose 
more or less of the menstruum, and, therefore, vary in composi- 
tion to a limited extent ; inal the definite hydrate can be obtained 
ee ey, repeat 
_ Another class of facts Sone on the same subject has been 
_ forced upon the during the last few years by the 
: Investigations he has undertaken on the revision of the atomic 
weights. In a memoir on the “Numerical Relations between 
can 
—as the resa 
had been cgi up to that time— 
S nite pr cote ia posi It tt is a course very easy to > dincover : 
oe ut how can you be sure that a given oxide, sulphide, ae 
 ¢hloride, iycaabis or iodide does not ~~ a similar co a 
of a lower or higher order? How ean you be sure that eit er - Le 
_ these compounds does not contain a went amount of oxide? — 
is such pr geet as these that are the chief source of aos 
tainty, and they cause variations of composition precisely sim 
_ lar to those discussed above. By referring to the examples cited 
on gd 65 of = volume | it will be found that they are. for the 
t part, if no © a suspicion of e of this ind, 
cod: 
