Dmitri Mendeleev noticed patterns in elements which allowed him to design the periodic table which ultimately led to the modern periodic table.

How were elements organized before this? Was there any method that served as a crude standard?

  • 2
    $\begingroup$ It just so happens that I'm reading a book that provides an awesome overview of the efforts to classify the elements. An answer summarizing this will not be quick to assemble, but I'll put it together when I can. (Unless someone gets Ihde's book and beats me to it, of course.) $\endgroup$
    – hBy2Py
    Commented Jul 17, 2017 at 0:25

2 Answers 2


Aaron J. Ihde's The Development of Modern Chemistry (DMC) has an entire chapter entitled "Classification of the Elements," which includes a very nice overview of the attempts before Mendeleev to bring order to the elements. All quotations below are from Chapter 9 (pp236-243, specifically) of the 1984 Dover edition of DMC, itself "an unabridged, slightly corrected republication of the third printing (1970) of the work originally published by Harper & Row, Publishers, Inc., in 1964" (front matter). The section and sub-section headers are generally those of DMC, with exceptions marked with an asterisk.

It is important to bear in mind when considering some of these systems that the best values of the atomic weights of many elements at the time were incorrect, either (a) due to simple experimental error, or (b) due to systematic misinterpretation of the available data, due to the limited state of knowledge at the time. For example, the best-regarded atomic weights of many elements were one-half the accepted values of today because the values were derived from measurements on compounds whose formulas were erroneous $($e.g., the atomic weight of oxygen was calculated based on experiments with water, which was thought to have the formula $\ce{HO})$.

For interest, a short article on Aaron J. Ihde can be found on the German language Wikipedia.

Early Attempts at Classification

Döbereiner's Triads

As already noted by Buttonwood in his answer, Johann Wolfgang Döbereiner appears to have been the first to attempt to systematize the elements, after:

he observed that the atomic weight of strontium appeard to be $50$, the mean of the then accepted values for calcium $(27.5)$ and barium $(72.5)$.


His triads included elements with similar properties, the atomic weight of the central member being the mean of the other two atomic weights.

The Döbereiner triads (some which he was unable to complete at the time) included in DMC (using modern atomic symbols) are:

$$ \require{begingroup}\begingroup \begin{array}{cccccccc} \hline \ce{Li} & \ce{Ca} & \ce{Cl} & \ce{S} & \ce{Mn} & \ce{B} & \ce{Be} & \ce{Y} \\ \ce{Na} & \ce{Sr} & \ce{Br} & \ce{Se} & \ce{Cr} & \ce{?} & \ce{?} & \ce{?} \\ \ce{K} & \ce{Ba} & \ce{I} & \ce{Te} & \ce{Fe} & \ce{Si} & \ce{Al} & \ce{Cs} \\ \hline \end{array} $$

DMC notes further:

Magnesium he considered as an isolated element, not part of a triad. Fluorine, as yet undiscovered but its existence clearly known, he did not include in the halogen triad. Although the atomic weight of nitrogen fell exactly between that of carbon and oxygen, Döbereiner regarded the three elements as isolated non-metals rather than as members of a triad. He also treated hydrogen as an isolated element.

Other Numerical Classification Systems

  • Similarity defined by an atomic weight difference of $8$:

    P. Kremers [no further identification given] observed that the atomic weights of certain [elements he regarded as] non-metals in which he saw similarity differed by $8$; i.e., $\ce{O} = 8$, $\ce{S} = 16$, $\ce{Ti} = 24.12$, $\ce{P} = 32$, $\ce{Se} = 39.62$.

  • Atomic weights as odd vs. even multiples of four:

    At the same time, [Kremers observed that] the atomic weights of certain metals fell between those of successive non-metals; i.e., $\ce{Mg} = 12$, $\ce{Ca} = 20$, $\ce{Fe} = 28$. When divided by $4$, the atomic weights of the non-metals were an even number, those of the metals an odd number.

  • Groupings of elements suggested by John Hall Gladstone based on (a) similar atomic weights, and atomic weights in (b) geometric and (c) arithmetic progression:

    • (a) $\ce{Cr/Mn/Fe/Co/Ni}$ all near $28$; $\ce{Pd/Rh/Ru}$ all near $52$

    • (b) $\ce{Pd}$ group $(52)$, $\ce{Pt}$ group $(99)$, and $\ce{Au}$ $(197)$; separated by multiplicative factors of approximately two

    • (c) $\ce{Li}$ $(7)$, $\ce{Na}$ $(23)$, and $\ce{K}$ $(39)$; separated by an additive factor of $16$

    The first type [Gladstone] compared with allotropy; the second, with polymerism in organic chemistry; the third, with homologous series.

  • Classifications based on arbitrary mathematical progressions:

    • Josiah Parsons Cooke (1827-1894) of Harvard developed a classification based upon six series of elements, each series derived from the atomic weight by a mathematical formula. For example, the sixth series, composed of hydrogen and the alkali metals, was based on the formula $1+(n\times 3)$. Thus: $$\begin{array}{ccc} \ce{H}~(1) = 1 + (0 \times 3) & ~~~ & \ce{Na}~(23) = 1 + (7 \times 3) \\ \ce{Li}~(7) = 1 + (2 \times 3) & ~~~ & ~\ce{K}~(39) = 1 + (13 \times 3) \end{array}$$

    • Jean Baptiste André Dumas (1800-1884) posited several arithmetic progressions in series of elements, roughly analogous to homologous series in organic compounds, in the vein of J.H. Gladstone's third type of grouping above (series names are my descriptors):

      $$\def\dumhc#1#2{\ce{#1} &= 1 + (#2 \times 14)} \textbf{Hydrocarbon Radicals} \\ \begin{align} \dumhc{H}{0} \\ \dumhc{CH3}{1} \\ \dumhc{C2H5}{2} \\ \dumhc{C3H7}{3} \\ \dumhc{C4H9}{4} \\ \textit{general}\dumhc{}{n} \end{align} $$ $$ \textbf{Halogens} \\ \begin{align} \ce{F} &= 19 \phantom{+2~~\times 16.5)+(2~~\times 28) + 19} = \,19 \\ \ce{Cl} &= 19 + \phantom{(2~~\times} 16.5 \phantom{)+(2~~\times 28)+19} = \,35.5 \\ \ce{Br} &= 19 + (2 \times 16.5) + \phantom{(2~~\times} 28\phantom{)~+19} = \, 80 \\ \ce{I} &= 19 + (2\times 16.5) + (2\times 28) + 19 = 127 \end{align} $$ $$ \textbf{Chalcogenides} \\ \begin{align} \ce{O} &= 8 \phantom{+(4~~\times 8)} = \,8 \\ \ce{S} &= 8 + \phantom{(4~~\times} 8\phantom{)} = 16 \\ \ce{Se} &= 8 + (4\times 8) = 40 \\ \ce{Te} &= 8 + (7\times 8) = 64 \end{align} $$ $$ \textbf{Pnictogens} \\ \begin{align} \ce{N} &= 14 \phantom{+17+(2~~\times 44)} = ~14 \\ \ce{P} &= 14 + 17 \phantom{+(2~~\times 44)} = ~31 \\ \ce{As} &= 14+17+ \phantom{(2~~\times} 44 \phantom{)} = ~75 \\ \ce{Sb} &= 14+17+(2\times 44) = 119 \\ \ce{Bi} &= 14+17+(4\times 44) = 207 \end{align} $$ $$ \textbf{Alkaline Earths}\textit{ (mostly)} \\ \begin{align} \ce{Mg} &= 12 \phantom{+(10~~\times 8)} = \,12 \\ \ce{Ca} &= 12 + \phantom{(10~~\times} 8\phantom{)} = \,20 \\ \ce{Sr} &= 12 + \phantom{1}(4\times 8) = \, 44 \\ \ce{Ba} &= 12 + \phantom{1}(7\times 8) = \, 68 \\ \ce{Pb} &= 24 + (10\times 8) = 104 \end{align} $$

      Interestingly, Dumas interpreted these "trends" as being potential evidence (obviously not borne out by subsequent investigation) for the feasibility of transmutation among specific groups of related elements.

Immediate Precursors of the Periodic Table

De Chancourtois' Telluric Helix

As noted by Buttonwood, Alexandre-Émile Béguyer de Chancourtois (1820-1886):

... conceived the idea of plotting the elements according to atomic weights on the surface of a cylinder. The circumference of the cylinder was divided into sixteen sections since the atomic weight of oxygen was $16$. The elements were plotted on a line or helix that descended at an angle of $45^\circ$ with the top of a cylinder. There was a striking resemblance in elements that were on the same vertical line.

... [H]e observed that atomic weights followed the formula $n+16n'$, the value of $n$ frequently being $7$ or $16$ [noting that none of the noble gases had yet been discovered]. ... Gaps in the helix were considered as indicating not unknown elements but different varieties of known elements. He believed, for example, that there was a form of carbon whose atomic weight was $44$.

In particular, the above Wikipedia article on de Chancourtois emphasizes that he "was the first scientist to see the periodicity of elements when they were arranged in order of their atomic weights." A public domain image of his "Telluric Helix" is also available at the Wikipedia page, and reproduced below (click to enlarge):

de Chancourtois classification scheme

Newlands' Law of Octaves

Again as noted by Buttonwood, John Alexander Reina Newlands (1837-1898):

... was the other major precursor of Mendeleev and Lothar Meyer. ... Newlands' first publication dealing with the classification of the elements appeared in 1863. It was essentially a reworking of Dumas' ideas, for Newlands was looking for numerical relationships between elements. A year later he adopted the atomic weights recommended by Cannizzaro and published a table of 37 elements subdivided into ten families. A crude repetitive character was evident, and blank spaces were left for undiscovered elements. In a table in a subsequent paper he assigned numbers to the elements [precursors to the modern atomic numbers], which were listed in order of increasing atomic weights. He used numbers for the elements in all his work thereafter. In August, 1865, he published an eight-column table listing 62 elements in order of increasing atomic weights; they were subdivided into seven horizontal families. It was in this table that he saw an analogy to the octave in music. The eighth element resembled the first element, the fifteenth resembled the first and the eighth; in other words, an interval of seven elements separated similar elements. Soon afterwards he suggested that the numerical relationships observed by earlier chemists were due to this law of octaves.


Newlands must be credited for taking a pioneering step toward the discovery of the periodic law, for he detected the repetition of properties when elements are arranged according to increasing atomic weights. He noted that this relation was evident only if Cannizzaro's atomic weights were used. he used blank spaces for unknown elements but failed to do this consistently, holding that perhaps the interval between repetitions was eight or nine and that this could be dealt with as new elements were discovered.

As with de Chancourtois, the Wikipedia article on Newlands includes a public domain image of his periodic chart, which is reproduced here for reference:

Newlands table of the elements



One attempt to order chemical elements was Döbereiner's system of triades, published in Annalen der Physik und Chemie, back in 1829 (doi 10.1002/andp.18290910217 with Wiley); or (open access with Gallica).

Although an actual view into his paper permits the speculation he was not using the atom masses we know today; rather than using specific weight and molecular weight as known / defined at his time of oxides of Ca, Ba, Sr, for example:

enter image description here

(The dots on top of Ca, Ba, Sr seem to adhere a nomenclature, as on top of S, Se, Te three of them; Cl, Br, I are topped by five.)

  • $\begingroup$ This is really interesting! Are there any other methods that you know of? $\endgroup$ Commented Jul 17, 2017 at 0:09
  • $\begingroup$ Beside the triades (groups of three), there was the attempt to order elements in groups of eight (de Chancourtois, en.wikipedia.org/wiki/…) and Newlands (en.wikipedia.org/wiki/John_Newlands_(chemist)). To know (and retain) about it was a bit by chance, as once a professor engaged in a dedicated subsection of a chemical society offered a weekly lecture "history of Chemistry" and a pinch of own curiosity "How did the ancestors before modern times?" similar to the one by ACS (scs.illinois.edu/~mainzv/HIST/Logo/logo.php). $\endgroup$
    – Buttonwood
    Commented Jul 17, 2017 at 21:18

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.