fulfil the conditions required. In accordance with generalisations which have been made correlating structure and properties, the first of these formulæ belongs to a primary alcohol, the second to a mixed ether: two, and only two compounds, C,H,O, are known, one exhibiting the properties of a primary alcohol, the other those of a mixed ether. When however n1 < 2n, + 2, and divalent atoms are also present in the molecule, the formula may contain only tetravalent carbon atoms, or it may contain tetravalent, and also dior trivalent carbon atoms. Thus in C,H,O n1 = 2n1; two structural formulæ are possible wherein each carbon atom is tetravalent, viz.

Each of these is the formula of an ether; in propylene oxide we have an ether the properties of which shew that it is probably described by the first of these formulæ. Six structural formulæ are possible for the molecule CHO, provided some of the carbon atoms may be tri- or divalent. Three compounds having this formula (besides propylene oxide) are known; of these, one is a ketone, i.e. belongs to a class of compounds the molecules of which are generally regarded as containing a carbon and an oxygen atom in direct union; another is an aldehyde, i.e. belongs to a class of compounds the molecules of which are regarded as containing a carbon atom in direct combination with one oxygen and one hydrogen atom; and the third is an alcohol, probably a primary alcohol. The six possible formulæ are

The first and second formulæ contain each one trivalent carbon atom, and the oxygen atom is monovalent in both, the third contains one trivalent, the fourth and fifth each one divalent carbon atom, and the sixth contains two trivalent carbon atoms. Formulæ (1) and (2) are appropriated by dimethyl ketone and propaldehyde respectively; of the remaining four, (3) and (5) represent allylic alcohol as a primary, (6) as a secondary, and (4) as a tertiary alcohol. Judging from the general reactions of allylic alcohol, this compound is probably a primary alcohol. Formula (3) is preferable to (5), because the latter would lead us to expect acetic acid /CH, as one of the

CO,H

products of oxidation of allylic alcohol; inasmuch as acetic acid is not produced in this oxidation, formula (3) more probably expresses the structure of the molecule of allylic alcohol than any other possible formula.

74. In these examples of the method adopted for determining the structural formula of a compound, several generalisations concerning the connection of structure with properties have been assumed: e.g. that, if a given compound exhibits aldehydic properties, the structural formula of the molecule is to be written as containing the atomic group COH; that two structures are possible for this group, one in which the carbon atom acts directly on the oxygen and on the hydrogen atoms (H-C-O), the other in which direct action occurs only between the carbon and the oxygen atoms (C-O-H); further, the first of these structures is assumed to be correlated with the group of properties connoted by the word 'aldehydic,' the second with the properties connoted by the expression 'tertiary alcoholic.' When therefore a new carbon compound is discovered, it is necessary to determine, as far as possible, to what group of compounds it belongs; the existence of a certain atomic group (or groups) in the molecule of the compound may then generally be predicated, and the number of possible structural formulæ may thus be considerably diminished. But the classification of the carbon compounds is certainly not yet complete;

hence arise two difficulties, (1) a new compound may belong to a class no other member of which has been previously examined, in which case no class-group can be assigned to the formula of the new compound; or (2) a compound may be prepared whose properties indicate that it belongs to one of the known classes, and yet the group which generally marks this class may not be present in the molecule of this particular compound. The following cases may be taken as illustrations of these difficulties.

(1) It was known that the action of nitrous acid on carbon compounds containing the group NH, (amido-derivatives) resulted in the production of compounds differing from the original by containing OH in place of NH,; but when nitrous acid acted on certain amido-derivatives of benzene, compound molecules containing one nitrogen atom more and two hydrogen atoms less than the original molecule were obtained. The reaction appeared to be abnormal. Several of the new compounds were prepared, their properties were studied, and the existence of a new class of carbon compounds was recognised, the relations of which to other classes appeared to be best summarised in formulæ containing the characteristic group - N,-.

Certain peculiar and definite properties appear to be always associated with this group; it has recently been shewn that the formation of molecules containing this group and derived from compounds of the 'fatty' or 'paraffinoid' series is possible under special conditions.

(2) As the result of long and varied experience, the generalisation has been made that the molecules of carbon acids contain the characteristic group H-O-C-O, but from time to time compounds have been prepared exhibiting acid properties, but possessed of a molecular structure from which the characteristic group is absent. Thus C.H yields C,H,NO,, and from this compound two isomerides. CH BRNO, are obtained, one of which is a monobasic acid, while the other does not shew acid properties; the possible formulæ for these isomerides are

(1) HC-C-CH3, and (2) H,C-C-C-NO2 ;

from a consideration of the general properties of the two isomerides and their relations to other compounds, the second formula is assigned to the acid. Hence we are obliged to conclude that although most known carbon acids are characterised by the atomic group H-O-C-O, yet a substance may be a true acid in the molecule of which this group is not present.

A very instructive example of the difficulties to be overcome before a general structural formula can be assigned to a group of carbon compounds, is afforded by the investigations which have been, and are being made into the constitution of the quinones1.

These examples (and others might easily be added) shew how undesirable it is to regard the present system of classification of carbon compounds as final. As facts are accumulated, the atomic grouping which was regarded as a classgroup sometimes becomes the group of a larger class, sub-classes being formed, each characterised by its special group, and yet each containing the class-group. Thus, from the analogy between metallic hydroxides and alcohols, and for other reasons, the group O-H was assigned to alcohols (e.g. C,H,. OH, C,H,. OH, &c., &c.) ; but it became evident that a sub-division of this great group was required; facts were amassed and formula devised to generalise these facts, until most chemists are now agreed that the molecules of those alcohols called 'primary' (which yield certain definite products when oxidised, &c.) contain the atomic group HO-CH,, the molecules of those called 'secondary' (and which yield other, but also definite products when oxidised) contain the group H-O-C – H, and the molecules of those called 'tertiary' (which yield a third distinct set of products when oxidised) contain the group C-O-H.

1 See Armstrong and Groves, loc. cit. pp. 812, 813.

Each of these 'alcoholic groups' itself contains the group O-H; but the 'acid group' H-O-C-O also contains this group; now we know that the function performed by hydrogen in an alcoholic molecule is not the same as that performed by hydrogen in an acid molecule,-e.g. all, or some of the hydrogen in the latter, but none of that in the former, is replaceable by metal when the compound is acted on by a metallic carbonate,-hence we infer that the function discharged by a given atom in a molecule depends not only on the nature of that atom, but also on the nature of the atoms with which it is directly, and indirectly, connected in the molecule.

In all the alcoholic groups (viz. H,C-OH, HC-OH, and C-OH) an atom of hydrogen is directly connected with an oxygen atom which is again connected with an atom of carbon, within the binding-sphere of which come either hydrogen atoms and atoms belonging to the other part of the molecule-always either carbon or hydrogen atoms or only the latter. In the acid group (O-C-OH) the carbon atom with which the hydrogen atom is indirectly connected (through an atom of oxygen) is itself directly connected with an oxygen atom, as well as with an atom, or atoms, belonging to the other part of the molecule. Now oxygen is a markedly electro-negative element; from the facts enumerated and from other similar facts, the generalisation has been made, that when an atom of hydrogen is within the binding-sphere of an atom of carbon, which also directly binds negative atoms, or negative groups of atoms, that hydrogen is, as a rule, 'replaceable by metal,' &c., i.e. that hydrogen fulfils the function of 'acid hydrogen'.'

1 I am aware that such expressions as are used in these paragraphs, 'a carbon atom is directly connected with,' or 'directly binds to itself, an atom of hydrogen ;' 'an atom of hydrogen comes within the binding-sphere of a carbon atom,' &c., are very easily misunderstood; they appear, at first sight, to convey much more precise information than they really do convey. I have more than once insisted on the importance of clearly remembering that these and similar expressions are attempts to summarise facts concerning the actions of compounds in terms of a special theory of the structure of compounds. Nor should it be forgotten that, granting the fundamental hypotheses of the molecular theory, and also granting