saa The Life and Work of Dimitri Ivanovich Mendeleev
San José State University
Thayer Watkins
Silicon Valley
& Tornado Alley

The Life and Work of
Dimitri Ivanovich Mendeléev

Dimitri Mendeleev

The Timeline of the Life of
Dimitri Mendeléev

Prior Formulation of Relationships
between Chemical Elements

John Dalton formulated the concept of atomic weights to go along with his idea of matter being made up of atoms. Johann Döberreiner and William Odling both attempted to create a system of relations but failed. Döberreiner however pointed out the existence of triads of elements such as Chlorine, Bromine and Iodine in which the atomic weight of the middle element is approximately the average of those of the two end elements. Lithium, sodium and potassium constitute another example of such a triad.

John Newlands in 1864-65 attempted the construction of an arrangement of elements but it involved so many misplacements and no provision for new discoveries that it was useless. The arrangement attempted by Odling suffered the same flaws.

Dimitri Mendeléev submitted the following article to Russian Chemical Society in March of 1869. This is the first publication of the Periodic Table of elements.

The Relations between the Properties
of Elements and their Atomic Weights

by Dimitri Ivanovich Mendeléev

The investigations regarding the simple relations of atomic weights have caused many, in particular, Dumas, Pettenkofer, Sokelow and others, to point out the numerical relations between the atomic weights of those elements which form a group; but, so far as I know, they have not led to a systematic arrangement of all known elements. I know only of an attempt by Lensson to satisfy this requirement that seems so natural. However, this system of triads of single bodies suffers from a certain ambiguity, since it possesses no definite principle as a basis. Lensson endeavors to support his classification of elements into triads with the help of the relations between atomic weights (in every triad the atomic weight of the in-between element is equal to half the sum of the atomic weights of the two outside elements, as was done first by Kremers and others); further, he claims support in chemical similarity and the colour of compounds. However, the classification according to the latter principle becomes uncertain as a result of the differences noted in the colors of Co--Cr--Cu--, and many other compounds according to the external condition they have undergone or according to the form of combination in which they are found. Yet there are natural groups to be found in Lensson's system which agree quite frequently with our general concepts, such as, e.g., K, Na, and Li; Ba, Sr, and Ca; Mg, Zn, and Cd; Ag, Pb, and Hg; S, Se, and Te; P, As,and Sb; Os, Pt, and Ir; Pd, Ru, and Rh; W, V, and Mo; Ta, Sn, and Ti, and others. But to place in one group Si, B and F; O, N, and C; Cr, Ni, and Cu; Be, Zn, and U as Lennson does, is hardly possible, after all. Moreover in his system the tendency appears to be implicit to subjugate the natural grouping of elements to the triads, which scarcely correspond to nature, and this is also not consonant with the certainty that the known series of elements are incomplete. If space could be found in his system to accommodate elements which are still to be discovered, this would result in the destruction of groups considered to be complete up to this time.

When I undertook to write a handbook of chemistry entitled "Foundations of Chemistry," I had to make a decision in favour of some system of elements in order not to be guided in their classification by accidental, or instinctive reasons, but by some exact, definite principle. In what has been said above we have seen the nearly complete absence of numerical relations in the construction of systems of elements; every system, however, that is based upon exactly observed numbers is to be preferred of course to other systems not based upon numbers because then only little margin is left to arbitrariness. The numerical data available regarding elements are limited at this time. Even if the physical properties of some of them have been determined accurately, this is true only of a very small number of elements. Properties, such as the optical and even the electrical or magnetic ones, cannot serve as basis for the system naturally, since one and the same body, according to the state in which it happens to be at the moment, may show enormous differences in this regard. With respect to this fact, it is sufficient to remember graphite and diamond, ordinary and red phosphorus. The vapour density which enables us to know the molecular weight of bodies is not only unknown for most elements but it is subject also to changes which agree completely with the polymetric transformations as they have been observed for compound bodies. Oxygen and sulfur furnish unambiguous proof for this fact; the relations between nitrogen, phosphorus and arsenic provide another confirmation, in so far as these similar elements possess the molecular weights N2, P4, As4, which are unequal to each other with respect to the number of atoms. But there is no doubt that the polymerization of an element must go hand in hand with the change of a number of its properties. One cannot be certain whether for any arbitrary chosen element, e.g., for platinum, another state would become known and that therefore, the place of a given element in the system would have to be changed according to its physical properties. However, everybody does understand that in all changes of properties of elements, something remains unchanged, and that when elements go into compounds this material something represents the [common] characteristics of compounds magnitude of the atomic weight, according to the actual, essential nature of the concept, is a quantity which does not refer to the momentary state of an element but belongs to a material part of it, a part which it has in common with the free element and with all its compounds. The atomic weight does not belong to coal and to the diamond but to carbon. The procedure, according to which Gerhardt and Cannizzaro have determined the atomic weights of elements is based upon such unshakeable and indubitable methods that for the majority of bodies and, in particular, for those elements whose heat capacity in the free state was already determined, there exist no longer any doubts about the atomic weight of the element. These doubts still existed a few years earlier when the atomic weight was so often confused with the equivalent weight and when it was determined according to different, even contradictory, principles.

For this reason I have endeavored to found the system upon the quantity of the atomic weight.

The first attempt I undertook in this direction was the following: I selected the bodies with the smallest atomic weight and ordered them according to the magnitude of their atomic weights. Thereby it appeared that there exists, a periodicty of properties and that even according to valency, one element follows the other in the order of an arithmetical sequence. [See Table 20-1.]

Table 20-1
Li = 7Be = 9.4B = 11C = 12N = 14O = 16F = 19
Na = 23Mg = 24 Al = 27.4Si = 28P = 31S = 32 Cl = 35.3
K = 39Ca = 40 Ti = 50V = 51

In the division of elements with an atomic weight greater than 100 we encounter a completely analogous series:

Ag = 108, Cd = 112, U = 116, Sn = 118,
Sb = 122, Te = 128, I = 127

It is seen that Li, Na, K, Ag show the same relationship to one another as do N, P, V, Sb, etc. Immediately the idea arose in me whether it was not possible to express the properties of elements by their atomic weights and whether one could not base a system upon this? In the following, the attempt as such a system is described.

In the proposed system, the atomic weight of an element serves to determine its place. Collecting the groups of elements known up to now, according to their atomic weight, leads to the conclusion that the method of ordering elements according to their atomic weight does not contrary, points directly toward it. In this regard the collection of the . . . six groups [in Table 20-2] is sufficient.

Table 20-2
 Ca = 40Sr = 87.6Ba = 137
Na = 23K = 39Rb = 85.4Cs = 133
F = 19Cl = 35.5Br = 80I = 127
O = 16S = 32Se =79.4Te = 128
N = 14P = 31As = 35Sb = 122
C = 12Si = 28 Sn = 118

These six groups show clearly that there exist certain, definite relations between the natural properties of elements and the magnitudes of their atomic weights. However, one should not imagine that such relations represent a picture of homology; this is not the case for the reason that, in those elements whose atomic weights are known with accuracy, no genuine homologous differences exist. Even though the difference in atomic weights of sodium and potassium, fluorine and chlorine, oxygen and sulfur, carbon and silicon amounts to 16, the difference between the atomic weights of nitrogen and phosphorus is 17, however, and�what is still more important�the differences between calcium and strontium, potassium and rubidium, chlorine and bromine, etc., are unequal; and the deviation in the first place, exhibits a certain regularity and, secondly it is much larger than the difference which could be ascribed to experimental error. In the collections indicated above the strict regularity is striking in the change of atomic weights within the horizontal rows and the vertical columns. Only the atomic weight of tellurium is out of place in the series; but this could easily be the case because it has not been determined accurately, and if we assume the atomic weight 126-124 instead of 128, then the system fits completely.

Thus the group of fluorine possesses elements which combine, preferentially, with a single atom of hydrogen, the group of oxygen with two, of nitrogen with three, of carbon with four atoms of hydrogen. Thus, in this respect, the naturalness of the group-classification, in an arrangement defined according to the numbers expressing the atomic weight, does not suffer any disturbance but, on the contrary, is suggested in advance.

In the first arrangement, we have 7 columns (perhaps the most natural ones also), of which Li and F are uni-valent and are most widely separated with respect to electrochemical behavior; Be and 0 which succeed them are bi-valent, then come B and N�tri-valent, and in the middle the quadri-valent carbon has its place. If we consider the distance of Na and Cl, Ag and I, and similar aspects, we notice that the arrangement of ele- tarn aegree to the valency analogous to the concept or similarity.

All comparisons carried out by me in this direction lead me to the conclusion that the magnitude of the atomic weight determines the properties and many reactions of a compound body. As soon as this assertion is verified in the further application of the proposed principle to the study of elements, then we shall approach the epoch where we understand conceptually the essential differences�and the reasons for the similarity, of the elements.

I state in advance that the law proposed by me does not contradict the general tendency in natural science and that, so far, it has not been proved, although suggestions of this kind did exist already. From now on, it appears to me, new interest will be awakened for the determination of atomic weights, for the discovery of new elements and for the finding of new analogies among the elements.

I shall now present one of the many systems of elements which are based upon the atomic weight. They form but one attempt to represent the results which can be achieved in this direction. I am quite conscious of the fact that this attempt is not final, but it appears to me to express quite clearly already the applicability of my proposed principle to all elements whose atomic weight is determined with some reliability. Above all I was interested to find out a general system of elements. The attempt is shown in Table 20-3.[This] table . . . has convinced me of the possibility

Table 20-3
Ti = 50Zr = 90? = 180
V = 51Nb = 94Ta =182
Cr = 52Mo = 96W = 186
Mn = 55Rh = 104.4Pt = 197.4
Fe = 56Ru = 104.4Ir = 198
Ni=Co = 59Pd = 106.6Os = 199
H = 1Cu = 63.4Ag = 108Hg = 200
Be = 9.4Mg = 24An = 66.2Cd = 112
B = 11Al = 27.4? = 68Ur = 116Au = 197?
C = 12Si = 28? = 70Sn = 118
N = 14P = 31As = 75Sb = 122Bi =
O = 16S = 32Se = 79.4Te = 128?
F = 19Cl = 35.5Br = 80I = 127
Li = 7Na = 23K = 39Rb = 85.4Cs = 133Ti = 204
Ca = 40Sr = 87.6Ba = 137 Pb = 207
? = 45Ce = 92
?Kr = 56La = 94
?Yt = 60Di = 95
?In = 75.6Th = 118

that the atomic weight of elements may be used as the basis of the system.* Initially I had ordered the elements into an uninterrupted sequence according to the magnitude of their atomic weights; but I noticed immediately that in the series of elements thus obtained some discontinuities are present. If one starts, e.g., with H = 1, then until Na = 23 there are present at least eight elements, and nearly the same number is found between elements with atomic weights of 23 and 56, 63 and 90, 100 and 140, 180 and 210, and in these particular groups of elements alone, the analogy is apparent by arranging them according to their atomic weights. In many cases there exist strong doubts still regarding the place of such elements which are not yet sufficiently investigated and which are placed close to the edges of the system as, e.g., Vanadium; according to the investigation of Roxoe it should be ascribed a place in the nitrogen group, but on the grounds of its atomic weight (51) it should be placed between phosphorus and arsenic. The physical properties equally support this position of vanadium; thus vanadium oxychioride, VOC13, is a liquid, having at 14° the specific weight of 1.841 and boiling at 127°, by which properties it approaches the corresponding phosphorus compound and places itself somewhat higher than the latter. If one assigns vanadium its place between phosphorus and arsenic, then we have to include in our table above a special column for vanadium and for the elements corresponding to it. In this column then, in the row containing carbon, a place will be opened up also for titanium. Titanium is related, according to this system, to silicon and tin just as vanadium is to phosphorus and antimony. Among these, in the succeeding row, which contains oxygen and sulfur, chromium is to be placed, perhaps; for chromium exhibits the same behavior toward sulfur and tellurium, as titanium does toward carbon and tin. In this case, we would have to place manganese, Mn = 55, between chlorine and bromine.

Thus, this part of [Table 20-3] would be composed as [shown in Table 20-4]. Evidently thereby, the natural connection between members of a horizontal row is disrupted, although manganese does show some similarity to chlorine, as does chromium to sulfur.

Table 20-4
Si = 28Ti = 50? = 70
P = 31V = 51As = 75
S = 32Cr = 52Se = 79
Cl = 35.5Mn = 55Br = 80

On top of this, however, it would be necessary to introduce another column between arsenic and antimony in order to include this group, columbium, Cb = 94, which represents the analogue of vanadium and antimony. Within the group magnesium, zinc and cadmium in this column, it appears that indium (In = 75.6) has to be placed provided it really belongs to this row (it is more volatile than Zn and Cd). Further, one would have to place zirconium in the row of carbon and tin, actually next to the latter, as the atomic weight of Zr is smaller than that of tin but larger than that of titanium. In this manner, there would be a vacant place in this horizontal row, for an element whose position would be between titanium and zirconium.

All the same, I decided not to construct the columns mentioned above, and this because analogies belonging indubitably to different rows would be left unconsidered. It suffices to point out that Mg, Zn, and Cd exhibit many analogies with Ca, Sr, and Ba, but to unite these bodies into one group: Mg = 24, Ca = 40, Zn = 65, Sr = 87.6, Cd = 112, Ba = 137, would, in my opinion, mean the destruction of the natural similarity of the elements.

For elements with small atomic weight, such as lithium and hydrogen, the first column is reserved, and in this way six columns would be obtained or eight, if we take special columns for Ti and Zr, over which all elements are distributed in horizontal rows whose numbers possess chemical similarity. Only the one row of lithium and sodium has representatives in all columns, the other rows have representatives only in a few columns, so that vacant places occur for elements which, perhaps, shall be discovered in the course of time. It must be remarked here that all elements occurring more frequently in nature possess atomic weights between 1 and 60, and these elements are

H, C, N, O, Na, Al, Fe, Ca, K, Cl, S, P, Si, Mg;

the higher atomic weights belong to elements that are rarely encountered in nature, which do not form large deposits and which, therefore, have been studied relatively little.

With respect to the position of some elements, there exists, quite understandably, complete uncertainty. In particular, this holds for those elements that are little investigated and whose atomic weight has hardly been determined correctly. To these elements belong, for example, yttrium, thorium, and indium.

It must be remarked, moreover, that the upper members of the fourth column (Mn, Fe, Co, Ni, Zn) form the transition to the lower members of the (third) column in which Ca, K, Cl and similar elements are found; thus, |cobalt and nickel, chromium, manganese, and iron represent, in their properties and with respect to atomic weight, the transition from copper and zinc to calcium and potassium. Perhaps their position may have to be changed for this reason; and if they were to be placed in the lower rows instead of the upper rows, then there would be three columns here which in many respects exhibit similarities; namely, one column comprising cobalt, nickel, chromium, manganese and iron, a second column with:�cerium, lanthanum and didymium, palladium, rhodium, ruthenium; finally a third column which contains platinum, iridium and osmium.

The system of elements proposed here is, of course, not to be considered as completely closed, but it appears to me to be based upon such data and such natural approximations, that its existence can hardly be regarded as doubtful; for the numbers confirm the similarities which result from the study of the compounds of the elements. A number of questions will arise when all elements are arranged into one whole, but the most interesting problem appears to me to be the arrangement of groups of similar elements, such as those of iron, cerium, palladium and platinum since, in this case, elements close to each other in their nature also exhibit approximately the same atomic weights, a circumstance not to be observed in other rows, for in the latter similar elements possess different atomic weights. It may be that the system of elements arranged in groups will, in consequence of the study of these groups, be changed in such a manner that in certain parts of the system, the similarity between members of horizontal rows has to be considered, but in other parts, the similarity between members of the vertical columns. In any event, it appears to be certain when we look at the proposed table, that in some rows the corresponding members are missing; this appears especially clearly, e.g., for the row of calcium; in which there are missing the members analogous to sodium and lithium; magnesium represents, to a certain degree, the analogue of sodium, but magnesium cannot be placed in the row of calcium, strontium and barium. This is proved not only by the properties of some compounds of these elements but also by those physical properties which are attributed to the metals themselves.

I cannot but direct attention to the fact that in the comparison of the lower members of the row with the upper there is noticeable a sharply distinct difference in properties and reactions. This is analogous to what we perceive in the series of organic homologues: in the upper members of the homologous series some of the characteristics belonging to the series are weakened; thus, for example, paraffin which was closed, at first, in the ethylene series, can be taken with the same (and naturally stronger) justification to belong to the series of marsh gas, since for such high homologues distinct characteristics cannot be expected in either this or that series. Similarly, characteristics of simple bodies, that show up strongly in the first column, become weakened in the last column formed by the heaviest elements. Lead, thallium, bismuth, gold, mercury, platinum, iridium. same time they are heavy elements, from which in many respects one single group could even be constructed without thereby destroying the foremost requirements of analogy. Thallium and bismuth are more widely separated in this relation, however, than lead and thallium, Jr bismuth and gold, mercury and platinum. At the same time, the elements standing lower than the halogen row possess oxides which exhibit basic properties rather than acid ones and which are the best representatives of the metals; while those elements which stand higher than the row of halogens possess either complete acid character or show transitional characteristics which lie between acid and base. For the latter reason also I could not persuade myself to put the iron group with the erbium-group in the lower part of the table.

Hydrogen has not yet been assigned a definite position because of its small atomic weight; it appears to me to be the most natural to place it in the row of copper, silver, and mercury, although it is possible that it belongs in some unknown row lower than that of copper. If it is permitted to express a wish when looking at the proposed [Table 20-3], then it is that the number of elements should be completed which stand closer to hydrogen. Those elements which form the transition from hydrogen to boron and carbon, would represent, naturally, the most important scientific achievement that may be expected upon acquaintance with bodies yet to be discovered. With regard to the bodies of the second [vertical] row it would be most promising, in my view, to subject beryllium and boron to an exact study and I shall endeavor to carry this out as soon as possible. In general, the elements with low atomic weight deserve greater scientific interest, to judge from what has been said previously, than those elements whose atomic weight is high. Considering the characteristics of the system, the remark has still to be made that some analogies are clearly visible from the table. Thus, C, B, Si, Al stand together, as well as Ba, Pb, Tl or V, Cr, Cb, Mo, Ta, W; others can be guessed in advance so to speak. Thus, there is no doubt that, for uranium, (but not for gold which, perhaps, has to be placed in the row of iron) a place has to be made in the row of boron and of aluminum, and, indeed, between these elements no little similarity exists. Thus, e.g., Turmeric is turned brown by the action of uranium oxide as well as by boric acid; the composition of borax, Na2B4O7, is analogous to that of the uranium compound K2U407. The compounds which aluminum oxide forms with a base have been little studied so far, and this question which has interested me already for some time, is going to form the subject-matter of one of my next communications.

In conclusion I do not deem it superfluous to summarize the results of what has been said above:

* Perhaps it may be more rational to arrange the proposed table in the following way:

then follows  CrMo  
but belowOSSeTe  

This would possess the advantage that elements, which are sharply distinguished as are Cl and Na, form the most outside rows, between which the elements have to be placed having a chemical character with less pronounced distinction. However, in this case the centre of the table would be nearly empty, and, moreover, it would be of a rather doubtful kind; while in the present arrangement the centre is indubitable and possesses many representatives; and all less known elements are placed above and below at the edges.

The two following sketches may show how diverse the arrangement can be to the fundamental principle stated in this essay.

BeMgCaZnSrCdBa Pb
BAl   U  Bi?
CSiTi ZrSn   
OS Se Te W 

Hereby the row Cr, Mn, Fe, Ni, Co must supply the transition (the atomic weights of 52-59) from the lower part of the 3rd column (where we have K, Ca, V) to the upper part of the 4th column (i.e. up to Cu); and, similarly, Mo, Rh, Ru, Pd form the transition from the 5th to the 6th column (up to Ag), as well as Au, Pt, Os, Ir, Hg from the 8th to the 9th column. In this manner a system of spiral form is obtained, in which the similarity in the number of alternate rows is particularly noticeable, f.i. in the second row is Be, Ca, Sr, Ba, Pb, and also Mg, Zn, Cd. The difference in atomic weights, in this case, is nearly equal for each vertical and horizontal row. If one separates from this system the members which resemble each other most, one obtains a system of the following type:

Above there will be
in the middle would stand
but below

Similar arrangements can be imagined in great numbers, but they do not change the essentials of the system.

The First Communication of Mendelée's
Periodic Law to the West

Mendeléev submitted an abstract of his above article to the German journal Zeitschrift für Chemie which became the first communication of his work to the West. Here is the abstract.

The World of the Atom makes this comment:

Many years later, Mendeleev found a difficulty in placing the elements of the argon group and radium, these substances having been discovered long subsequently to the formulation of the "periodic" scheme. In an article written for the "Russian Encyclopaedia," and abstracted into English (Nature ," November, 1904), he later acknowledges the independent existence of these elements, and places the argon group in a column by themselves. The first place in the same column is assigned to the ether, which he assumed to be molecular in structure with a very small atomic weight. (page 301)


Henry A. Boorse and Lloyd Motz (eds.) The World of the Atom, Basic Books Inc., New York, 1966.

HOME PAGE OF applet-magic