MELLOR and Maley1'2 have recently pointed out that the stability of complexes of bivalent metal ions follows the order Pd>Cu>Ni>Co>Zn>Cd>Fe>Mn>Mg 0 --1Mn irrespective of the nature of the ligands involved, and remark that "although it is not clear what determines the order of metals in the series, it is worth noting that metals forming dsps bonds (Cu11 and Pd11) head the list". Although our own programme of experimental work is incomplete, it appears desirable to indicate some of our conclusions in this field, since they extend the observations of Mellor and Maley and render to some extent more intelligible the order of decreasing stability found experimentally.The numbera give the value of n in log kn for: o, ammonia; O, ethylene diammine; x, propylene diammine; 9, salicyl- aldehyde If published data2 on the successive stability constants of divalent ions of the first transition series are plotted against their atomic number (see graph), it is seen that the stability of complexes increases steadily to reach a maximum at copper whether the ligands be ammonia, ethylene diammine, propylene diammine or salicylaldehyde. The reluctance of copper to exceed a co-ordination number of 4 is shown by the low value of kt for the pentammino-, and k3 for the tris-ethylenediammino-eomplexes. All zinc complexes are markedly less stable than those of Cu11, and it is perhaps significant that the M-shell is completely filled in Zn11, whereas the attachment of six groups through d2sp3 or dsp2 orbitals (involving resonance) is possible throughout the sequence Ca11 to Cu11. A similar trend of stabilities can be noted in the second transition series with Pd*l>Cdn, but the interposition of Ag11 (which forms complexes isomorphous with those of Cu11) can as yet be only inferred. In the third transition series, the order of stability is certainly PtH>HgU and Pb11. Published data on the absolute stability of complexes are limited so far to those already quoted. Yet it is obvious that the formation of an inner complex represented by the equationM "+ + n HB ^ MBn + nH+ involves essentially a competition between (hydrated) protons and metal cations for the anion of the reagent forming the complex: the more stable the complex, the lower the pS. at which it can persist, and vice versa. (It is, of course, implicit that any ligand displaces an equivalent number of water molecules from the hydrated cation; for example, ammonia or pyridine displace one; diammine, salicylaldehyde or dithizone displace two, etc.) It can be shown that the pR for 50 per cent precipitation (or 50 per cent extraction of metal complexes for a constant excess of reagent) is a measure of their relative stability. Such data, with figures for the pH. of incipient precipitation of metal quinaldinates and, for comparison, the stability constants of salicylaldehyde complexes are collected in the accompanying table and confirm the sequences of stability already noted.Metal 'Oxine'; j>H for 50% precipitation" 'Dithizone'; pB. for 50% extraction with carbon tetrachloride" Quinaldinic acid; pJl for incipient precipitation4 Salicylalde- hyde; log k^kz in aqueous dioxane1'2 Mn 4-8 ~9 6-8Fe - 5-6 - 7-6 Co 3-2 4-9 3-9 8-3Ni 3-2 3-0 3-9 9-2 Cu 2-3 1-0 0-5 13-3Zn 3-1 3-2 1-85 8-1 Pd ~-2-0 2-5 14-8Cd . 4-2 4-1 3-2 7-8 Pt ~-2-0Hg - ~-2-0 0-2 6-8 Pb 5-2 5-5 3-6 9-1It is worth noting that among the divalent cations of the first transition series, copper alone is precipitated by 5: 7-dibrom-8-hydroxyquinoline in normal acid5, and similar examples of the outstanding stability of cupric complexes are commonplaces of analytical chemistry. H. Fischer6 noted a similar order of stability for complexes with oe-nitroso-(3-naphthol, (3-nitroso-oe-naphthol and diphenyl carb-azone, though the positions of cobalt and nickel were sometimes interchanged. It must be emphasized that the order may be radically changed if the divalency of any of the participants is altered. Thus the selectivity of nitrosonaphthols as reagents for cobalt depends on their capacity to oxidize it and form a complex of trivalent cobalt far stabler than the corresponding complexes of divalent iron and nickel (or cobalt). That divalent copper, palladium (and trivalent iron) interfere with the cobalt test7 shows that their complexes are comparable in stability even With those of trivalent cobalt. Derangement of the 'natural' order of stability may also arise from steric and other factors. Thus, copper forms normal 1: 2 complexes with M-methyl- and NN'-diethyl-ethylenediammine but cannot do so with N-methyl (or ethyl)-N'-diethyl-ethyienediarnmine8; and trivalent gallium, indium and thallium, but not aluminium, will give insoluble 1: 3 complexes with 2-methyl-8-hydroxyquinoline and similar reagents9. We shall present elsewhere our observations with sterically hindered ligands such as a-methyl- and a: a'-dimethyl-o-phenanthroline.The stability of complexes formed by any one ligand, for example, ammonia, with a series of metals may be expected to increase with the electro-negativity of the metal concerned. Though reliable values for the transition elements are not yet available, it is of interest to record that a plot of instability constants (log Tcn) against the second ionization potential (corresponding to the change M ->M++-\- 2e for the metal concerned) is approximately linear from manganese to copper, points for zinc lying somewhat off the curves. Thence, or less readily from the accompanying graph, it can be seen that the successive increments in stability in passing from manganese to copper complexes are of much the same order whether water is replaced by two molecules of ammonia, or one of salicylaldehyde or a diammine. When any one metal is considered, the gain in configurational entropy consequent upon ring formation is strikingly demonstrated by the high relative stability of the latter complexes. Whereas it seems reasonably certain that the most stable complexes are always formed by elements terminating the transition series, and probable that the same 'natural' order of stability is preserved throughout each series, the acceptance of a unique order of stability operating throughout the Periodic Table must wait upon more extensive quantitative measurements with more varied types of ligands and upon a wider range of metals. Quantitative studies on chromous complexes should prove of particular interest.