Af>rt/2\, 1887] 



NA TURE 



597 



noticeable, viz. that the mtitals in them apparently retain the 

 properties which they exhibited in the parent atomic compounds. 

 Everyone knows the marked difl'erence in properties of ferrous as 

 contrasted with ferric salts : they differ not only in chemical 

 behaviour, but also in their physical properties, and are readily 

 distinguishable by their colour. The properties of the ferrous 

 molecular compounds, however, are those of the simple fer- 

 rous compounds : ferrous potassium chloride, for example, 

 FcoClj . C1.,K„, is a green salt much like ferrous sulphate. 

 Facts such as these have led me to suggest that in such cases 

 the formation of the molecular compi'itnd is due to the attraction 

 of the negative element of the one ''atomic " compound by the 

 negative element of the other, the metal having no influence 

 except that the amount of affinity of which the negative element 

 is possessed depends on the nature of the metal with which it is 

 associated. It would in fact appear that hydrogen and the 

 metals generally may be regarded as the analogues of the 

 C„Ho„ + i and C^H^^-- hydrocarbon radicles, and that their 

 compounds with negative elements may be likened to un- 

 saturated hydrocarbons of the form C„H.,„ + [ . CH . CHj. We 

 know that whenever such a compound etiters into combination, 

 the C„Ho„4i radicle takes no part in the change, combination 

 of whatever kind being effected by means of the unsaturated 

 radicle, CH . CH„, with which it is associated. I do not mean 

 to contend that the metals are fully neutralised in their com- 

 pounds, but merely that as a rule they behave as though they 

 were saturated, just as do the C„Ho- - radicles derived from the 

 benzenes. There can be little doubt that an absolute distinction 

 must be drawn between hydrogen and the metals on the one 

 hand, and the non-metals on the other. Regarding the facts in 

 the light of our knowledge of carbon compounds, it is difficult 

 to resist the conclusion that the differences observed are due to 

 differences in structure of the stuffs of which the elements as 

 we know them are composed, the which differences condition 

 perhaps a different distribution of the electric charge or its 

 equivalent, in the case of each element. 



Addendum, April 18S7. — I will row venture to call attention 

 to the points which after a year's further consideration of the 

 subject appear to me of special importance. 



We are as far as we ever were from being able to define a 

 "simple electrolyte " in the chemical sense — that is to say, to 

 define the class or classes of compounds to which simple electro- 

 lytes belong. The investigation of the electrical behaviour of 

 ttirc compounds is therefore of the highest importance ; it is 

 essential, however, to bear in mind that not only must pure 

 compounds be studied, but scrupulous care must be taken to 

 guard against a possible decomposition of the substance under 

 examination, either by heat alone, or by contact with the 

 electrodes or the containing vessel. I believe that the con- 

 clusions which Clark based on his most interesting observations 

 on the electrolysis of mercuric salts are vitiated by some such 

 effect having been overlooked. The experimental difficulties 

 surrounding the problem are therefore veiy great ; and the more 

 hopeful method of attacking it in many cases would appear to 

 be that adopted in Kohlrausch's experiments on the specific 

 resistance of water : in other words, to determine the influence of 

 impurities. 



A similar problem relates to the possibility of basing a de- 

 finition of a non-metal as distinct from a metal on electrical 

 properties. It is well known that no consistent definition can be 

 given, and that we are at present obliged to base our division of 

 the elements into metals and non-metals on general considerations. 

 Now, although metals differ enormously in specific resistance, 

 the metals as a class oppose a comparatively feeble resistance to 

 the passage of electricity, and moreover resistance always in- 

 creases as the temperature of a metal rises ; it is therefore note- 

 worthy that not only is the specific resistance of non-metals, 

 such as carbon, phosphorus, selenium and sulphur, enormously 

 great in comparison with that of metals, but that it dimini-hes 

 as temperature rises : non-metals therefore behave in this respect 

 as electrolytes, and as no special precautions have hitherto been 

 taken to obtain pure non-metals for the purpose it is well worth 

 while to ascertain if the specific resistance offered by non-metals 

 be not the greater the nearer the approximation to purity. 



To determine the valency or atom-fixing power of an element, 

 according to present views, it is necessary to determine the 

 number of atoms which can enter into direct association with an 

 atom of the element considered ; and this necessarily involves a 

 discussion of the nature of " atomic " as distinct'' from "mole- 



cular " compounds. The electrical hypothesis that an atom of 

 unit valency carries unit charge, a dyad two such charges, a 

 triad three, involves the more specific determination of the 

 number of charges which are associated with any particular 

 atom ; but, again, on this hypothesis we have to determine 

 whether any real distinction can be drawn between atomic and 

 molecular compounds, and whether an atom having, say, unit 

 charge, has the power of combining with more than a single atom. 

 My own view certainly is that atomic and molecular compounds 

 are specifically distinct ; and that in the latter the number of 

 atoms associated with what may be regarded as the grouping 

 element or elements in the compound is in excess of the number 

 of unit charges which the particular element or elements of 

 necessity carry. Taking nitrogen as an example, it appears to 

 me that the whole of the evidence to be derived from the study of 

 nitrogen compounds is compatible with the assumption that nitro- 

 gen carries at most three charges : that it is a triad, in fact ; and I 

 am inclined to regard the ammonium compounds of the type 

 NH3 . HX' as molecular compounds in which the residual affinities 

 of N and X' serve to unite H3N with X'H. The more I study the 

 question the more I incline to the belief that sooner or later we 

 must accept Kekule's ruling, unfashionable as this has become of 

 late years. 



The foregoing may appear to many to be but a restatement of 

 the tenets of the advocates of the doctrine of fixed valency. It 

 appears to me, however, that in the hypothesis of a definite 

 unit charge — in other words, of a definite unit valency — we 

 have a conception which for the first time enables us to frame a 

 consistent doctrine of valency : a given atom may be assumed 

 to carry under all circumstances a certain defi-nite charge, and 

 the problem which the chemist has to determine is, firstly, the 

 number of unit charges associated with any particular atom, and, 

 secondly, the manner in which the charge is, as it were, dis- 

 tributed when the atom is brought into association with other 

 atoms. The following illustration will perhaps serve to make 

 my meaning clear. Let unit charge or unit valency be regarded 

 as a unit "line of affinity" passing through the atom, and let it 

 be supposed that the atom moves upon this line of affinity with 

 a degree of freedom depending on its nature ; then it may be 

 supposed that combination between two atoms consists in the 

 overlapping of the lines of affinity. If each atom move out to 

 the end of its line of affinity, no part of the line will overhang ; 

 the molecule will consequently be saturated ; there will be no 

 residual affinity. On this view the stability of a molecule will 

 depend on the extent to which the lines of affinity of the con- 

 stituent atoms overlap, and its tendency to associate with other 

 molecules will depend on the extent to which the lines of 

 affinity of the constituent atoms overhang. For example, it 

 may be supposed that in ordinary chlorine the two atoms of 

 which the molecule consists have moved out near to the end of 

 the lines of affinity, so that there is but little overhang : the 

 molecule is therefore of considemble stability, but forms un- 

 stable combinations with other molecules ; in iodine, on the 

 other hand, it may be supposed that the atoms are closer to- 

 gether, the lines of affinity overlapping less than those of 

 chlorine ; consequently the molecule is less stable than that of 

 chlorine, but may form more stable molecular compounds as the 

 lines of affinity overhang to a greater extent than do those of 

 chlorine. It is possible in this way to understand that an atom 

 which carries but unit charge — a monad, in fact — may enter into 

 association with two distinct atoms. Adhering to the above 

 symbolic language, it may even be suggested that perhaps the 

 difference between a non-metal and a metal may be that the 

 structure of the non-metals is such that they move with difficulty 

 upon their lines of affinity, and probably in a very limited 

 number of directions, and with unequal freedom in different 

 directions ; and that the structure of metals is such that they 

 move with comparative freedom upon their lines of affinity, in 

 some cases even with complete freedom, and almost equally so 

 in several directions. 



Speculations such as these are of value only if they serve as 

 a guide to further inquiry. I venture to put them forward in 

 the hope of inducing chemists to devote more attention to the 

 study of molecular compounds, for it is in this direction that we 

 are likely to gather most important information as to the valency 

 of elements other than carbon and hydrogen. Of late years such 

 inquiries have been but rarely pursued, and no doubt they are 

 less attractive than those which result in some new synthesis or 

 the determination of the constitution of an organic product ; but 

 their future value will be great, and the number of workers is 



