Validity, and Residual Affinity. 301 



historic interest in the discussion, though scarcely consistent. 

 Prof. Armstrong more logically inclines to the acceptance of 

 only one number for valency concerned with what may be 

 called chemical affinity of the first magnitude, and suggests 

 that besides this affinity proper (say a trivalent atomic charge) 

 there are residual affinities in play which probably account 

 for molecular compounds and aggregates. Curiously enough 

 he would, in many cases, thus place among " molecular com- 

 pounds " many hitherto generally recognized ''chemical 

 molecules." Chemists will be loth to part with these. It is, 

 as it were, taking our friends' dogs to the dog's home when 

 the street dogs really demand our attention. He supposes 

 that NMe 4 I is "the molecular compound" (tetramethyl 

 ammonium iodide) caused by the union of the two molecules 

 Me 3 N and IMe determined by the " residual affinities " of 

 the N and I atoms represented thus : — 



H 3 C. 



h 3 c _ ^Vr T— — ch 3 . 



H 3 C 



Whilst agreeing with the associations thus indicated, I propose 

 that the N and I atoms be named in such cases " quinqui valid " 

 and "tervalid " respectively, 



H 3 C X 



H 3 C-N V == I 111 — CH 3 ; 



H,(/ 



leaving it open for future research and discussion to determine 

 the exact nature and limits of validity and valency, since we 

 cannot at present commit ourselves to any partial theory, say 

 either that of Helmholtz or of J. J. Thomson. Indeed we 

 have also as chemists to keep an open mind on the sphere 

 and vortex-ring hypotheses of an atom *. It may, however, 



* Professor Armstrong says, " It may well be argued that we have no 

 reason to assume that nitrogen is pentad in the ammonium compounds, 

 or phosphorus pentad in the phosphoniimi compounds, or sulphur tetrad 

 in the sulphine compounds " (p. 24). My proposal involves the idea 

 that nitrogen is pentad in whatever sense chlorine is triad in NH 4 C1 taken 

 as H 3 N= =01 — H, there being no reason to assume that the four atoms 

 of hydrogen are directly associated with the nitrogen. Similarly for 

 phosphonium compounds ; hut for other reasons (e. g. the evidence of the 

 phosphihic acids, of PO(C 2 H 5 ) 3 , PO(C 8 H g ) 3 , and other bodies) the penta- 

 valency of phosphorus seems nmch better established than that of nitrogen. 

 This observation tallys with the recognition of an " allotropic form " of 



