Why the Periodic Table Is Shaped Like That

Look at the periodic table as a piece of graphic design and it is a mess. The first row has two entries with a canyon of white space between them. The next two rows are eight wide. Then it suddenly swells to eighteen for two rows. Then two more rows would be thirty-two wide, except that somebody has cut a strip out of each and parked the two strips underneath, separated from the body by a gap, usually with a little marker where they were removed.

No designer would produce that. And yet every serious version of the chart has that outline, because the outline was not chosen. Those row lengths — 2, 8, 8, 18, 18, 32, 32 — are a consequence of something that has nothing to do with charts at all, and the person who first drew the table had no idea what it was.

The row lengths are a list of four numbers

Here are the numbers that matter, and there are only four of them. An electron subshell is labelled s, p, d or f, and each type holds a fixed maximum number of electrons:

• an s subshell holds 2
• a p subshell holds 6
• a d subshell holds 10
• an f subshell holds 14

Now add them up in order:

• 2
• 2 + 6 = 8
• 2 + 6 + 10 = 18
• 2 + 6 + 10 + 14 = 32

Those are the row lengths. Not approximately — exactly, in order, with nothing left over. The periodic table is eighteen columns wide in its usual form because 2 + 6 + 10 is eighteen, and the four blocks of the chart are literally named after the subshells whose widths they are. The s, p, d and f blocks are two, six, ten and fourteen columns across respectively, and a row of the table is one complete pass through whichever subshells happen to be available.

Those capacities come from the quantum numbers, and specifically from the number of distinct orbitals of each type: one s orbital, three p, five d, seven f, each holding two electrons of opposite spin. That is where 2, 6, 10 and 14 come from, and the shape of every orbital and the counting behind it is the subject of its own guide.

Why each length appears twice

If the row lengths were just cumulative capacities you would expect 2, 8, 18, 32 — four rows, each longer than the last. Instead you get each number twice over. That doubling is the second half of the shape, and it comes from the order in which subshells fill.

Electrons do not fill by shell number. They fill roughly in order of energy, and by the time you reach the fourth shell the energies have started to overlap: the 4s subshell sits below the 3d. So the fourth row of the table starts with two s-block entries, then does the 3d subshell it skipped, then finishes with the 4p. Eighteen wide. The fifth row repeats the pattern one shell up — 5s, 4d, 5p — and is also eighteen wide. The same thing happens again with f: the sixth row runs 6s, 4f, 5d, 6p and comes to thirty-two, and the seventh row runs 7s, 5f, 6d, 7p and comes to thirty-two as well.

That overlap is usually taught as the diagonal rule, or the n+ℓ rule, and it is the single fact that makes a filling order writable. Working through it properly — including why the rule is a good approximation rather than a law, and the score of elements whose real configurations break it — belongs on the page for writing an electron configuration, and there is a calculator that will produce any element's configuration with those exceptions flagged.

The exceptions are worth knowing exist, because they show the ordering is a tendency and not a decree. Chromium and copper both promote an electron out of the 4s to give a half-filled or filled d subshell, and neither of them moves anywhere on the chart as a result. Position on the table is set by proton count, not by whichever configuration an atom settles into.

The hole in the first row, and the missing d-block in the second

Two of the table's oddities are just the capacity arithmetic running out of subshells.

The first shell has only an s subshell. There is no 1p — the rules that generate orbitals do not allow a p subshell until the second shell — so period 1 has exactly two entries. Hydrogen and helium sit at opposite ends of a row that is drawn eighteen columns wide, which is why the top of the chart looks broken. It is not broken. There is genuinely nothing that could go between them.

The second shell has s and p but no d, for the same reason one shell down. So period 2 is eight wide, and the transition metals do not appear until period 4, where the 3d subshell finally becomes accessible. That is the whole explanation for the notch in the upper middle of the chart. It is a hole in the table because there is a hole in the physics.

The two floating rows are a paper-size decision

Here is the part that genuinely is a human choice, and the only part.

The full-width table is thirty-two columns across. It exists, it is correct, and it is sometimes printed. But at thirty-two columns each cell has to shrink by nearly half to fit the same page, and almost nobody needs to read the f-block on a wall chart at a glance. So the conventional layout cuts the fourteen f-block columns out of periods 6 and 7 and lays them below the main body as two strips: the lanthanides and the actinides.

Two consequences follow, and both confuse people who were never told the strips had been moved. The strips are usually printed fifteen entries wide rather than fourteen, because the element at the far end of each is conventionally included with them. And the gap between the strips and the body is pure typography — those elements are not chemically detached from anything, they are sitting in the middle of their own rows with the ends of the rows printed above them.

Splitting the actinides out as a row of their own was itself an argument. Before the 1940s they were commonly slotted into the transition metals, and the case for treating them as an f-block series parallel to the lanthanides was made by Glenn Seaborg during the Manhattan Project. He was reportedly advised that publishing it would destroy his scientific reputation. It did not.

There are layouts that refuse the compromise entirely. The left-step arrangement published by Charles Janet in 1928, derived from a helix wound on nested cylinders, reorders the blocks so that the f-block comes first and the s-block sits on the right, producing a shape whose row lengths are 2, 2, 8, 8, 18, 18, 32, 32 — perfectly regular, doubling in clean pairs, and far closer to the filling order than the standard chart. It is arguably the more honest diagram. It is also nearly unusable for looking up a halogen, which is what most people open a periodic table to do.

Mendeleev built the right shape from the wrong information

Everything above was unavailable in 1869. The electron was not identified until 1897, the nuclear atom not until 1911, and the quantum account of subshells not until the 1920s. Dmitri Mendeleev had atomic weights and chemical behaviour, and that was the entire toolkit.

He ordered the known elements by weight and noticed that chemical properties recurred at intervals. Then he did the thing that made him famous: where the recurrence did not line up, he assumed the elements were missing rather than that the pattern was wrong, and left the cells empty.

In his 1871 revision he went further and predicted what would fill three of the gaps, naming them after the element above each hole:

• Eka-aluminium — predicted atomic mass around 68, density around 6.0, a low melting point, an oxide of the form R₂O₃. Gallium was found in 1875 by Paul-Émile Lecoq de Boisbaudran, mass 69.72, oxide Ga₂O₃, and it melts in a warm room. Lecoq's first density measurement came out at 4.7; Mendeleev wrote to tell him the value was wrong and should be nearer 5.9. Lecoq repurified his sample, measured again, and got 5.9.
• Eka-boron — predicted mass around 44. Scandium was isolated in 1879 by Lars Fredrik Nilson, who was not looking for it and did not recognise what he had; his Uppsala colleague Per Teodor Cleve made the identification with Mendeleev's prediction. Its standard atomic weight is 44.96.
• Eka-silicon — predicted mass around 72, density around 5.5, a high melting point and a refractory dioxide. Germanium turned up in 1886 in a silver ore from Freiberg, isolated by Clemens Winkler, at mass 72.63 and density 5.32. Winkler initially thought he had eka-antimony and had to be talked out of it.

Three predictions, three hits, one of them accurate enough to correct the discoverer's own laboratory work. That is why the table was accepted so quickly by people who had no idea why it worked.

The places where weight and chemistry disagreed

Mendeleev's ordering principle also broke in three specific places, and how he handled them says a lot about what the table actually encodes.

Tellurium is heavier than iodine, but tellurium behaves like selenium and iodine behaves like bromine, so Mendeleev put them in chemical order and assumed the weights were mismeasured. The same inversion occurs with argon and potassium, and again with cobalt and nickel. In all three cases the atomic weights were fine and the ordering principle was slightly wrong.

Henry Moseley resolved it in 1913 by measuring the X-ray spectra of the elements and finding that the frequency scaled cleanly with a whole number that stepped by exactly one from element to element. That number was the nuclear charge — the proton count. Ordering by atomic number rather than atomic weight fixes all three inversions at once, because atomic weight includes neutrons and the neutron count wanders. The gap between an element's atomic number and its atomic weight is a fact about isotopes, which is a different subject that happens to sit under the same table.

Moseley also settled how many elements were still missing between hydrogen and uranium, turning an open-ended search into a finite list. He was killed at Gallipoli two years later, aged 27.

One thing Mendeleev's table did not have was a column for the noble gases. Argon was not isolated until 1894, and the group had to be added wholesale as a new column on the right — a genuine structural amendment to a table that was already famous, and one it absorbed without difficulty.

What the columns are called, and the argument about one of them

The column numbering is the least interesting part of the chart and has caused the most trouble. For most of the twentieth century two incompatible schemes were in use: a European convention and a Chemical Abstracts Service convention, both of which labelled columns with Roman numerals and an A or B suffix, and which assigned those suffixes to different columns. A table marked "group IIIB" meant one thing in Leeds and another in Chicago. IUPAC settled it in 1988 by numbering the columns 1 to 18, left to right, with no letters. That is why old textbooks and new ones disagree, and why the modern numbering carries no chemical information whatsoever — it is a column index.

Which does not mean every column is settled. Group 3 is genuinely disputed. Everyone agrees it starts with scandium and yttrium; the question is what sits below them. One convention continues with lanthanum and actinium, the other with lutetium and lawrencium, and the two produce different-looking charts. IUPAC convened a task group under Eric Scerri to rule on it; its provisional report, published in Chemistry International in 2021, concluded that there is no objective basis for deciding between them — the matter is one of convention, which is exactly why a ruling is needed. None has been issued.

Helium has a similar problem in miniature. Its configuration is 1s², which by the block logic above puts it in the s-block, above beryllium. Its chemistry is that of an utterly unreactive gas, which puts it above neon in group 18. Standard tables choose chemistry; Janet's chooses configuration. Both are defensible and the disagreement has run for the best part of a century.

Where the bottom edge is

The seventh row was completed in 2016, when the last four gaps were filled and named: elements 113, 115, 117 and 118. Oganesson closes the row at the bottom-right corner, and the story of how those superheavy elements were made and claimed is a fight in its own right.

An eighth row would begin with a g-block, since the g subshell holds eighteen. That row would be fifty entries wide, which would make the wide-form table unprintable and the conventional layout even more of a compromise than it already is. Whether it will ever be populated is a question about nuclear stability rather than about chemistry, and nobody has made element 119.

What the outline is a picture of

The periodic table's outline is a picture of subshell capacity. Rows are 2, 8, 18 and 32 long because subshells hold 2, 6, 10 and 14 electrons and those are the running totals; each length appears twice because the 4s fills before the 3d and the same overlap repeats one shell up; the first row is short and the second has a notch because the p and d subshells do not exist that early; and the f-block hangs below the table for no better reason than that thirty-two columns will not fit on a page.

Mendeleev derived that shape without knowing any of it, from weights and chemical behaviour alone, and it held up because weight and proton count run in nearly the same order. Where they disagreed he trusted the chemistry and was right three times. If you want the practical version of all this — what the position of a cell tells you about the element in it — start with reading the table and the trends across it, or go straight to the elements themselves.