The limits of Mendeleev’s insight

2019 has been declared the ‘Year of the Periodic Table’, but I do not believe that the sesquicentenary of Mendeleev’s version of February 1869 is a cause for celebration. It was not an advance on Ernst Meyer’s version of 1868, mislaid by his publisher, nor even on William Odling’s of 1864.^[1] The big breakthrough came in the 22 months that led to Mendeleev’s table of November 1870, with the detailed prediction of eka-boron, eka-aluminium and eka-silicon.

The table of 1869 presented the elements in six unnumbered columns, the future rows, in order of increasing atomic weight (with the exception of iodine and tellurium, the atomic weight of which he queried in accordance with the chemical behaviour). The rows, the future groups, were also not numbered. The first column contained only hydrogen at the top and lithium at the bottom, but Mendeleev expressed the wish, not the prediction, that ‘the number of elements near hydrogen should be filled in [forming] the transition from hydrogen to boron and carbon, [which] would represent the most important scientific achievement to be expected…’.^[2] The second column contained the future ‘typical elements’ of groups II to VII plus sodium at the end aligned with lithium.^[3] Columns three to six foreshadowed the future alternation of rows with VII groups and rows with VIII groups. ‘?=50’, ‘?=68’ and ‘?=70’, the future ‘eka-boron’, ‘eka-aluminium’ and ‘eka-silicon’ – were in the appropriate places, but as yet they had no names nor any prediction of their properties. Five rare earths plus indium and thorium, with incorrect atomic weights, were at the bottom of columns three and four (Figure 1).

Figure 1:

Table of 1869.

In the 22 months to November 1870, Mendeleev made a number of very significant changes. First, he turned the columns into rows and vice versa, which made the table much easier to read, as the atomic weights now ran in the usual reading direction. It is possible that the new layout was inspired by the way the Devanagari script is laid out, with rows that represent the initial consonants and columns that show the combinations of these consonants with successive vowels. Mendeleev was friendly with a Sanskrit scholar at the time.^[4] The columns, numbered I to VIII, represented groups of elements united by their highest oxidation state, indicated by a generalized oxide formula at the top^[5] or bottom. This contradicted what he had said in 1869: ‘Oxygen compounds cannot serve as a guiding principle for the determination of the valences of the elements’.^[6] Within each of columns I to VII there is a left-side and a right-side sub-column. As there has been confusion over the use of ‘a’ and ‘b’ to distinguish these, I shall label them with subscripts _L and _R. There is some mystery over group VIII, in which the only element then known to have a potential valence of eight was osmium. Mendeleev referred to the elements of this group as a ‘transition series’, because they seem to transition back down towards a valency of one or two.^[7]

The first two rows are not numbered, so the numbering begins with the third row. The top row contains only hydrogen, still with no predicted elements to its right. Lithium moves to head the seven ‘typical elements’ in the next row, in groups I_L to VII_L (fluorine). The basic organization – in alternating rows of six (or seven) elements in groups VII R groups and 12 (or 11) elements in VII L groups plus group VIII – begins only in the third row, numbered row 1, which is offset in relation to the ‘typical elements, despite their great similarities. Two rows make one ‘period’, so rows 1–10 form the first to fifth periods. His first period, rows numbered 1 and 2, comprises sodium to chlorine and potassium to copper, but copper is optionally shifted down to appear in brackets in group I_R in row 3. The same pattern is repeated in the second period, rows 3 and 4, (copper or) zinc to silver, with silver optionally shifted down. The system worked in his first two periods, thanks to the happy fact that it corresponds with what we now see as the 18 groups of the (s+p+d) blocks with double counting of the ‘coinage metals compensating for the missing noble gases.

The VII/VIII system breaks down in the third and fourth periods, between the middle of row 6 and the beginning of row 8, with row 7 completely empty. This is where Mendeleev should have placed the rare earths, of which six had been named by 1870: cerium (1803), lanthanum (1839), and ‘didymium’ (1843), all found in ceria; also yttrium (1794), erbium and terbium (1843), all found in yttria. Apart from yttrium, they had the right atomic weights to be placed in the gap between his third and fourth periods; instead, he predicted non-existent analogues of the elements in rows 1–5. The VII/VIII system seemed to resume between tantalum in row 8 and thorium and uranium in row 10, which he placed in groups IV_L and VI_L – an understandable error, which persisted for many years, even into Bohr’s table of 1922. This was a misfortune, because it hardened Mendeleev’s commitment to his system. Was this alternation of VII and VIII groups what Mendeleev meant by his ‘Periodic Law’? If so, did it not trouble him that hydrogen and the ‘typical elements’ were outside the system? (Figure 2).

Figure 2:

Table of 1870.

It is remarkable that Mendeleev was attached to Bunsen’s laboratory in Heidelberg from April 1859 to February 1861 without becoming interested in spectroscopy, which Kirchoff developed in that place and at that time, using it to discover caesium and rubidium. Apparently, Mendeleev found the fumes and noise of the laboratory so annoying that he transformed his apartment into a laboratory with its own gas supply, where he conducted experiments on capillarity and alcohol solutions.^[8] He thus cut himself off from one of the greatest advances in chemistry and denied himself the possibility of sharing in the spectroscopic discovery of a dozen new elements, including gallium – one of his predictions.

What made Mendeleev’s enduring reputation was his successful prediction of the properties of three missing elements, for which he used the Sanskrit words eka, dvi, tri – one two three. Eka-aluminium (gallium) and eka-silicon (germanium) are indeed p-block analogues of aluminium and silicon, and the predictions were well founded, though he did not foresee the extraordinarily low melting point of metallic gallium, but then who could have done? However, eka-boron – scandium – was problematic; being the first element in what we now see as the d-block, it did not have a lower analogue. In fact, without the displacement of the ‘typical elements’, aluminium would have been the true eka-boron. Mendeleev’s predictions for scandium were saved by the fact that he based them hardly at all on the properties of boron but on those of its neighbours, calcium and titanium, and its true congeners, aluminium and yttrium. So much for the predictive value of his L and R groups!

The next 25 years were an extraordinarily fruitful period for science, with the discovery of van der Waals forces, the experiment of Michelson and Morley, the discovery of X-rays and radioactivity, Thomson’s plum-pudding model of the atom, the discovery of the electron, and Einstein’s explanation of the Brownian motion. However, Mendeleev’s final version of his table, from the 8^th and last edition of Principles of Chemistry (1906) shows hardly any advance on that of 1870, and indeed some regression. He had added scandium, gallium, germanium and radium, (but not polonium or actinium), and the noble gases, which form sub-column 0_L of a new group zero, but with neon in 0_R. Hydrogen is no longer alone in the top row, where a space is marked for element zero, presumably the ether (in spite of Michelson-Morley), and six spaces on the right (as wished for, but not predicted, in 1869). By 1906, all the ‘rare earth’ elements had been discovered, except lutecium (and promethium), but only lanthanum, cerium and ytterbium are entered, with 17 other spaces marked, including the whole of what is now row 9. Copper, silver and gold have been definitively moved to group I_R under hydrogen and sodium, and the brackets that had initially enclosed them there have been moved to their mention in group VIII (Figure 3).

Figure 3:

Table of 1906.

The overall structure of the table has changed subtly. The numbering of rows now starts with hydrogen, so that there are 12 instead of 10, and the term ‘period’ has disappeared from the margin. The lines across the table have altered the pairing of rows, so that the formerly ‘typical’ elements are coupled with the next row, but still with the same offsetting, with helium to fluorine in L groups and neon to chlorine in R groups. Starting with potassium, the alternation of rows seems to have become VIII/VII instead of VII/VIII. It is difficult to know what significance to attach to this, but it makes periodicity look more anomalous than ever, since now there are three rows of VII at the top. Essentially what we have in the 1906 version is the same as in 1869; perhaps its very fixity contributed to the way it continued to be the standard in education until the 1930s. Other authors were not so conservative. Henry Bassett in 1892 published a table with ten lanthanides separated from the main blocks. He placed thorium and uranium parallel with cerium and praseodymium, respectively, anticipating the actinide series. Alfred Werner’s 1905 table was equally modern, looking remarkably like the conventional table of today, although he was less prescient with the sequence of the actinides. Also, there were still helical versions.

Although he was happy to take credit for his predicted elements, Mendeleev lost interest in his table years before they were discovered. He wanted something precise like Newton’s laws of motion but he could not find any mathematical expression for his ‘Periodic Law’. In 1872 he turned to physics and launched a major research programme into the gas laws in the Russian Technical Society, in the vain hope of finding the ether. If he had not been so narrowly focussed, he might have discovered argon. Indeed, I have suggested elsewhere that if he had conceived the Periodic System as a spiral, he might have noticed that between the halogens and the alkali metals, there were gaps in the atomic numbers wide enough to accommodate elements with zero valency.^[9] His gas work occupied him until the end of 1880, though he found time to attack spiritualism, which was increasingly fashionable in St Petersburg.

Mendeleev’s rejection by the Russian Academy of Sciences in 1880 was a heavy blow. It came at a time when he was in love with a young art student and negotiating a tricky divorce. He even tried to reinvent himself as an art critic, and he launched into a career as an economist, also finding time to go up in a hot-air balloon to observe the total solar eclipse of 1887. From this he turned to the organization of education in the Empire and moved on to the general reform of the government. His life-long advocacy of the metric system saw him appointed Director of the Chief Bureau of Weights and Measures in 1893, and he was the main author of the law of 1899, defining all other Russian systems in metric terms. There seemed to be no limit to his intellectual and social ambition.^[10] No wonder he was not thinking about his periodic table! He had a brief revival of interest when it was suggested he might be candidate for a Nobel Prize, and in 1904 he published an ill-judged attempt to include the ether in his table, in spite of Michelson-Morley.^[11] Of course it was to no avail; by the terms of the bequest, the prizes were for those who ‘during the preceding year, have conferred the greatest benefit to humankind’.

A number of things limited Mendeleev’s ability to progress in chemistry after 1871. His insatiable intellectual interests and his ambitious nature drove him repeatedly into new fields of activity. A certain rigidity of mind prevented him from re-examining old assumptions, and it did not help that his table met with widespread adulation, and that the works of Chancourtois, Odling, Newlands and Meyer were virtually unknown. His achievement in 1870 was considerable, but he might have contributed so much more. It is essentially a sad story.

Note: Articles by Mendeleev are quoted in the English version and with the page numbers of William B. Jensen’s edition, referenced below.