ON THE MOLECULE OF WATER. 
179 
taining but one proportion of ethyl and 4*7 parts of nitrogen would bear no relat ion of 
series to the boiling-points and vapour-densities of bodies containing three proporti ons of 
conjoint hydrogen and ethyl united with fourteen parts by weight of nitrogen. More¬ 
over, by adopting the trihydric model for ammonia, corresponding to the formula H 3 N"', 
we perceive at once why the replacement of its hydrogen must take place in thirds ; or, 
in other words, why we cannot obtain ammonias in which one-half or one-fourth of the 
hydrogen is replaced by equivalent substitution, but only those in which one-third or 
two-thirds is so replaced; thus :— 
Again, ammonias are known to chemists in which two-thirds of the hydrogen are re¬ 
placed by one, and the remaining third of hydrogen by some other radicle ; or in which 
all three thirds are replaced by three different radicles, as instanced by diethyl-phenyla- 
mine Et 2 PhN'", and methyl-ethyl-phenylamine MeEtPhN'", for example. 
Seeing, then, that there are bodies, such as marsh-gas and silicated hydrogen, of which 
the hydrogen is replaceable in fourths at four successive stages, and which are conse¬ 
quently represented with four atoms of hydrogen; and that there are other bodies, such 
as ammonia and phosphoretted hydrogen, of which the hydrogen is replaceable in thirds 
at three successive stages, and which are consequently represented with three atoms of 
hydrogen,—it is not a priori improbable that there may be yet other bodies, such as 
water and sulphydric acid, in which the hydrogen is replaceable in halves at two succes¬ 
sive stages, and which must consequently be represented with two atoms of hydrogen. 
If we act upon water by excess of metallic sodium, for instance, we obtain the compound 
known as oxide of sodium ; and accordingly as we represent the molecule of water by 
the formula HO', in which O' stands for eight parts by weight of oxygen, or by the for¬ 
mula H 2 0", in which 0" stands for sixteen parts by weight of oxygen, so must we repre¬ 
sent oxide of sodium by the formula NaO', or Na. 2 0"; thus :— 
O' = 8 
(8) H O' 
(16) H O'. Na O' 
(8) Na O' 
0" = 1G 
Ho 0" (16) 
Na H" 0" (16) 
Nao 0" (16) 
But exactly intermediate between water and oxide of sodium, we have the very definite 
compound known as hydrate of sodium, which consists of one combining proportion of 
sodium and one combining proportion of hydrogen united with sixteen parts by weight 
of oxygen, and which cannot be represented save with sixteen parts by weight of oxygen. 
Now, just as the three intermediate bodies between marsh-gas and chloride of carbon 
force us to represent these extreme bodies with twelve parts of carbon, so that all five 
may appear as members of one and the same series; just as the two intermediate bodies 
between ammonia and triethylamine force us to represent these extreme bodies with 
fourteen parts of nitrogen, so that all four may appear as members of one and the same 
series; so must the intermediate body between water and oxide of sodium force us to 
represent these extreme bodies with sixteen parts of oxygen, so that all three may appear 
as members of one and the same series. If marsh-gas and chloride of carbon must be 
represented by formulae similar to those of their intermediate bodies, and consequently 
with four atoms of hydrogen and chlorine respectively; if ammonia and triethylamine 
must be represented by formulae analogous to those of their intermediate bodies, and con¬ 
sequently with three atoms of hydrogen and of ethyl respectively; then water and oxide 
of sodium must be represented by formulae analogous to that of their intermediate body, 
and consequently with two atoms of hydrogen and of sodium respectively. In the inter¬ 
mediate carbon-bodies, the remaining and replaced hydrogen form fourths of the whole ; 
in the intermediate nitrogen bodies, the remaining and replaced hydrogen form thirds of 
the whole; while in the intermediate oxygen compound, the remaining and replaced 
hydrogen form halves of the whole. Hence if the molecule of marsh-gas is to be repre- 
