Nobel Lecture, December 6, 1919

Atomic Weights

With deep earnestness I express my heartfelt gratitude for the great honour which you, eminent Fellows of this celebrated Royal Academy, have bestowed upon me. This honour gave me pleasure which could not be overestimated or over-stated.

No chemist, and especially no chemist who has endeavored to determine atomic weights, could come hither without feeling, in addition to his respect for the present Academy, a profound reverence for its past. Here in Stockholm the great Berzelius made his home; and of this Academy he was for years Permanent Secretary. As long as civilization persists his name will be reckoned among those of the great leaders of chemistry. Few men have been able to make so great an advance beyond their predecessors as that accomplished by this wise and energetic student of Nature. His many discoveries and his patiently executed and amazingly accurate revisions of Dalton’s atomic weights have exercised a lasting effect on science. His commanding sway for years over chemical thought throughout the world marks him indisputably as a thinker of the highest order.

Your present lecturer is a modest follower in Berzelius’s footsteps. The trail which Berzelius opened, and which others have cleared, made possible further advance. Every student of Science, even if he cannot start his journey where his predecessors left off, can at least travel their beaten track more quickly than they could while they were clearing the way: and so before his race is run he comes to virgin forest and becomes himself a pioneer. Ungrateful would he be who failed to remember his debt to his predecessors. It is an especial pleasure to acknowledge my debt to Berzelius in this his native country, as well as to admit my obligation to de Marignac, Stas, and the others who have done so much to advance the study of atomic weights.

The investigations now to be brought to your attention were first inspired by an intense philosophic desire to know more about the fundamental nature of matter, and its relations to energy. Later I perceived more and more clearly that a better understanding of the behavior of matter – the carrier of those complex chemical changes called Life – must give mankind more power over the circumstances of living. In short then, scientific curiosity on the one hand, and the hope of providing a basis for bettering man’s condition on the other, were the leading impulses of the work.

The special field chosen included the study of the fundamental properties of the chemical elements and of their simpler compounds in relation to one another. This study was begin with the hope that if man could only understand more about the elements- those beginnings of all earthly things – further light might develop in all of the complicated phenomena which ultimately depend upon their properties. Hence the work to be described was part of a comprehensive scheme involving the study of the most important attributes of the chemical elements.

In this day of scientific upheaval, one may well ask: “What are the chemical elements?” Are we to assume with recent theorists that all matter is composed of nothing but electrons and positive nuclei consisting perhaps of helium, because the radioactive elements seem partly to split into these components? Or are the old chemical elements of one’s youth still worthy of recognition? Both of these questions may be answered in the affirmative without inconsistency; for even if uranium splits apart in successive stages with the production of helium, a number of electrons and a kind of lead, and even if radium is nothing but a transition stage in this disintegration, nevertheless the ordinary elements are clearly no less permanent to-day than they were fifty years ago. Even supposing that the common elements, too, may be suffering disintegration over a period of countless eons, this decomposition is so exceedingly slow that in all the ordinary concerns of humanity these common elements may be considered as permanent. However they may be composed, and to whatever extent their structure may suffer change under extraordinary or abnormal stress, they are for practical purposes the obvious basis of our reasoning about the chemical mechanism of Life. Moreover, we must remember that only the very heaviest of the elements with atomic weights above 210 give definite evidence of disintegration, whereas, on the other hand, all the elements which are known to be essential to our living on earth have atomic weights of less than 56. Among these only potassium shows any considerable sign of radioactivity, and in its atoms as yet no sign of actual loss of weight has been detected.

Hence the properties of these old-fashioned but nevertheless still abiding chemical elements seem worthy of the most searching study in order to determine: in the first place what in each case the facts may be, and in the next place what correlation may exist among the facts.

Among the many data concerning the elements, the first which I selected for study were the atomic weights. These unique numbers seemed to be the most definite and unchanging of all the properties of material. They represent not only the relative weights in which substances as they exist on earth combine with one another, but also, because of the apparently exact parallelism between the mysterious force of gravitation and the cryptic tendency of inertia, the relative masses in which these elements combine. We cannot help believing that they give us at least a picture of the average relative weights of the atoms; and because they represent simultaneously two properties – gravitation and mass – they claim especial attention. Moreover, they seem to possess an extraordinary degree of constancy and definiteness. So far as we know, the law of combining proportions is one of the very few precise laws of the universe, complicated to an extraordinarily small degree by superposed irregularities. Importance and definiteness thus together offer a great temptation to the investigator who wishes to find out more about the nature of material. Until we understand the significance of these weights, we cannot hope to understand the atoms themselves, nor all the countless substances which are made from them. Far deeper in meaning than the accidental astronomical constants, such as the length of the day or the length of the year, they seem to stand out as the peculiar and basic attributes of those ninety or more primordial substances of which everything is built.

If our inconceivably ancient Universe even had any beginning, the conditions determining that beginning must even now be engraved in the atomic weights. They are the hieroglyphics which tell in a language of their own the story of the birth or evolution of all matter, and the Periodic Table containing the classification of the elements is the Rosetta Stone which may enable us to interpret them. Until, however, these hieroglyphics are clearly visible in their true form, we cannot hope for an interpretation. The first task of the investigator is to define sharply the outlines of these graven characters, in order that their true form may be manifest. Then perhaps there is hope of deciphering their meaning.

The importance of accurate knowledge in a case of this sort was foreseen long ago by Plato, who perhaps drew his inspiration from yet more ancient knowledge, coming from wise men of the Far East. As I have often quoted, he said: “If from any art that which concerns weighing and measuring and arithmetic is taken away, how little is left of that art!” The implication of this wise saying as regards the study of atomic weights is clear; any increase in the accuracy of the determination of these quantities must of necessity add greatly to our insight into the profound mysteries with which chemistry has to deal.

Accordingly, the atomic weights were chosen as the first property of material to be investigated. As time went on it became more and more evident that for a deep understanding of the atomic weights themselves one must likewise study other properties; and in recent years, although the atomic weights have been by no means forsaken, much time has been devoted to the study of compressibility and many other fundamental attributes of the elements. About these other researches, however, it is not my purpose to speak to you to-day. Indeed, since many of them are far from finished, it would be premature to dwell upon them now. To-day I call your attention only to atomic weights.

Let us turn first to the consideration of the actual task of determination. Experimental work of great refinement is necessary in order to determine atomic weights. No relationships between them have yet been certainly found which make it possible for us to compute by any sort of calculation exactly the value of any one atomic weight from any other. We must find by actual experiment the amount of one element which actually combines with the given amount of some other element, producing a pure compound of definite composition.

The experimental work usually resolves itself naturally in several successive processes. In the first place, substances to be weighed must all be capable of actual isolation in a pure state, uncontaminated by any kind of admixture. This is no easy task. Whether we weigh the elements in their uncombined state or weigh them in the form of some compound of known composition, we must be very sure that conditions are such as to make possible the exclusion of all complicating impurities from the scale pan.

Thus it comes to pass that comparatively few compounds of any given element are fit to serve as a means of determining its atomic weight, for the reason that comparatively few substances may be prepared in a perfectly pure state. The choice of the compounds to be employed is in some ways the most crucial part of the whole process, for with some compounds no result worthy of consideration could be obtained, even using the greatest care possible. To repeat, then, the first task is the choice of materials to be employed.

The second task is a corollary of the first. Having chosen wisely, the experimenter must then prepare the substances, whatever they may be, in a state of the greatest possible purity. He must never forget that every precipitate carries down with it contaminating impurities adsorbed or included by the substance as it separates from the solution. He must remember always that no receptacle necessary to contain the substance is free from the possibility of being dissolved and thus affecting the result. Moreover, precipitates are never wholly insoluble; and most substances will volatilize and lose some of their weight if heated to an excessive temperature. These complicating circumstances combine often in unexpected ways to introduce impurity into one’s preparation, and the experimenter must not only guard against these dangers, but must prove by adequate and satisfactory tests that no such complication has occurred. Moreover, above all he must not forget that oxygen, nitrogen and water are almost omnipresent, and continual care must be exercised lest in some way one of these impurities may affect the substance which is serving as the basis of the work.

These difficulties which hamper progress are serious enough even in the first part of the preparation, during which the substance needed to serve as the starting part of the research is made ready; but they are multiplied tenfold or one hundredfold during the latter part of the work. This is because during the first part of the work the hampering influences are mitigated by the circumstance that much of the material may be sacrificed during purification, whereas after the beginning of the quantitative experiment, not only must the substance be kept in a pure state, but also it must be collected to the last trace and brought on to the balance pan. If any, even a tenth of a milligram, escapes collection, the loss must be estimated by careful experiments, so that its exact amount may be known. In this work, as at a trial in court, the witness must testify as to the whole truth, and nothing but the truth. Such an investigation, to have merit, must be conducted with ceaseless attention to these rules of procedure. I have tried always to be sure that the substance being weighed represented all the substance which I was seeking to weigh, and nothing more; and whenever possible I have not contented myself with a hypothetical presumption that such was the case, but have endeavoured to prove by special experiments, first, that nothing was lost, and, secondly, that no foreign substance had been unintentionally included. Usually, if the experimenter gives this matter sufficient thought, and if he is sufficiently impressed with the importance of certainty on these points, a fairly satisfactory proof is obtainable. He does well to discover such a proof before publication, and not to leave the matter to the subsequent investigation of others.

The investigation which introduced me to the study of atomic weights was a comparison of the atomic weights of oxygen and hydrogen. This investigation, in which I assisted Professor Josiah Parsons Cooke of Harvard University (where nearly all of my work has been done) was the first to bring direct evidence that the ratio between these atomic weights is considerably less than 16 to 1. We weighed hydrogen gas in globes, properly counterpoised, and then burned it to water by means of copper oxide, weighing the product. After two years of work with hydrogen of great purity prepared by several methods, we came to the conclusion that the atomic weight of hydrogen must be not far from 1.008 if oxygen is taken as 16.000. The present accepted value, due to E.W. Morley and Lord Rayleigh, which has been confirmed in many ways, is very near this figure. The subject is one of great importance, because so many other atomic weights are referred directly to oxygen. The training acquired in this work was of great value, and from that time forward was used and developed in independent investigations.

The next interesting question, undertaken in part simultaneously with this, was the study of the atomic weight of copper. The material was taken from two widely separate parts of the world, in order to determine if different specimens of copper carefully purified give the same values. The primary object of the choice of two sources of copper was the detection of possible unknown impurities, but the inquiry was in a sense a prevision of the extraordinary phenomenon recently demonstrated in the case of lead. That there might be different kinds of copper seemed to me not altogether outside the range of possibility. The result, published in 1887, was, however, in this respect entirely negative. Copper from America gave the same result as that from Germany. The important outcome of the research was the discovery that the accepted atomic weight of copper was nearly half a per cent too low.

This outcome was a revelation to me as to the inaccuracy of some of the earlier work which appeared on its face to be satisfactory. Naturally it gave great incentive to study other atomic weights, in order to discover if in these also grave errors existed. The first undertaking was barium, because in the course of the copper work, the precipitation of barium sulphate from copper sulphate seemed to show that barium too had been incorrectly determined. Later, because of the advantage to theory of comparing elements in the same families of the Periodic System, the other metals of the alkaline earths were likewise studied by means of the same methods, in the (as yet vain) hope of finding exact mathematical relationships.

While working with the atomic weight of strontium in 1894, the first form of a simple, but highly useful apparatus was devised for enclosing and weighing hygroscopic substances out of contact with the laboratory air – an apparatus which had much to do with subsequent success. This was later modified for cases demanding ignition under more varied circumstances, in a research on the atomic weight of magnesium in 1896. In this latter research, as well as in most of the subsequent ones, I have had the good fortune to be assisted by able young collaborators -for the most part graduate students in Harvard University – to whom I take great pleasure in acknowledging my indebtedness. Without their help the range and quantity of the researches must have been very much less. Generous grants from the Carnegie Institution of Washington since 1902 have likewise greatly helped the work. So, too, have the recent building and endowment of the admirable Wolcott Gibbs Memorial Laboratory by friends of Harvard University interested in these investigations.

One of the next problems of interest was a pair of parallel series of experiments upon cobalt and nickel, undertaken in order to determine whether or not these two elements have the same atomic weight, as some persons thought at that time. The antithesis between the problem concerning copper (wherein the question was as to the possibility of a single element’s having two atomic weights), and that of nickel and cobalt (wherein the question was as to the possibility of two elements’ having the same atomic weight), was interesting and suggestive. Exactly the same methods were used for nickel and cobalt, both of the metals being converted into bromide with pure bromine, the bromide being analyzed for its content of halogen and of metal. The result proved beyond cavil that cobalt really has a higher atomic weight than nickel – a result of much interest, since the properties of the two elements, as well as their atomic numbers recently determined by X-ray spectra, suggest the contrary sequence.

This outcome added to my conviction that the table of the Periodic System represents only in a very crude fashion relationships which are really highly complex and subtle. Clearly the nature of the elements is not always capable of being depicted by any such simple sequence of atomic weights and properties as may be ordered by placing the elements in definite pigeonholes. This complexity of behavior is especially true of the elements with atomic weights over that of potassium. Another important aspect of the nickel and cobalt analyses lay in the fact that here, for the first time among the investigations now being described, a complete analysis was effected. Both of the factors in a binary compound were determined. Because the total result added up very nearly to one hundred per cent, confidence in the earlier work was engendered, and more security for the future was felt.

Next I turned to the interesting question as to the relation between atomic weights and Faraday’s electrochemical equivalents. Lord Rayleigh had long previously shown by a very carefully conducted and important research that a fairly close agreement exists between these quantities, but the full measure of modern exactness had not been attained. A careful study of the silver and copper coulometers, in which, of course, these two metals respectively are precipitated by the passage of a galvanic current, revealed several errors in each. In the first place, it was found that, because of the reducing action of metallic copper on cupric sulphate to form cuprous sulphate, too little copper was deposited in the copper coulometer to correspond to the current. By using different-sized cathodes and extrapolating to zero-surface, this error was eliminated. On the other hand, it was found that in the silver coulometer when the anode is wrapped with filter paper (as it always was in the earlier researches), too much silver is deposited, or at least the total weight of the precipitate is too large. By using a porous cup the contaminations from the anode were eliminated, and a much more accurate value was found.

With these precautions, the weight of silver deposited in the silver coulometer was found to bear the same relation to the weight of copper deposited in the copper coulometer as the atomic weight of silver bears to one-half the new atomic weight of copper, within a reasonable limit of error. Thus it was clear that within this limit of error Faraday’s law held true. In order to discover if change of temperature and solvent had any effect, the weight of silver deposited at 250° from a solution of silver nitrate in fused sodium and potassium nitrates was compared with that of silver deposited by the same current from an aqueous solution; and the two were found to be identical. Thus there seemed to be but little doubt that Faraday’s law is to be counted as among the most precise of all the laws of Nature. Deviations noticeable in other cases are probably to be ascribed simply to superposed side reactions like that which somewhat complicates the case of copper.

Even before the study of radioactivity great interest was attached to uranium, on account of its unique position at the end of the system of atomic weights. The thought was pertinent that possibly this element with a high atomic weight might exhibit irregularities in its quantitative behavior not shown by the others; and the consequent curiosity was greatly heightened by Becquerel’s and Madame Curie‘s brilliant discovery. The study of uranium was, therefore, begun in 1896, and extended over a period of four years.

Many compounds were investigated at first with unsatisfactory results, usually due to the manifold valences of the element. After much trouble and thought the tetrabromide was selected as the typical compound to be analyzed, for reasons too lengthy to be discussed here. Various difficulties too numerous to mention were at least partially overcome, and a fairly satisfactory series of results, pointing to the preliminary value of 238.5 (instead of the old value 240) were obtained. The hope that this research, even if not final, should be a help to others, has been justified very recently – for Hönigschmid, repeating the same method with all the details of most recent Harvard practice and with the advantage (not available in 1900) of quartz apparatus, has found the value 238.2, only about 0.12% lower.

Passing over the perhaps no less interesting cases of strontium, zinc, magnesium, iron and caesium, because their lesson is not unlike that of those just described, we come to a crucial series of experiments, namely, that concerning the atomic weights of sodium and chlorine, which was begun in 1903.

Although Stas, who had studied these two elements with great care, was an excellent experimenter, and although he added much to our knowledge of method, and contributed highly valuable results, nevertheless he was but human, and, as the sequel proved, he had the misfortune to overlook certain important sources of error in his work. The discovery came to me in an interesting fashion. Sodium bromide had been prepared in a state of very great purity for the purpose of determining the transition temperature of its hydrated crystals. Such a transition temperature, especially of a substance with small heat of transition, is very sensitive to impurity. This sodium bromide, which gave a very constant transition point, was analyzed with precision as a matter of routine. To my amazement more bromine was found in it than corresponded to the atomic weights as determined by Stas. The only reasonable explanation seemed to be that Stas’s atomic weight of sodium was too high. Such an iconoclastic conclusion, however, needed verification in other ways. Especially Stas’s experiments with common salt must be repeated, for sodium chloride was the chief substance upon which his atomic weight of sodium rested. The research was not easy, and because it involved the disproving of results among the most carefully obtained in the whole field of chemistry, it demanded unusual precautions and meticulous care. The story is long, and can only be briefly summarized here. We found that not only the atomic weight of sodium, but also that of chlorine, was in error. Indeed, the error in the latter was partly responsible for that in the former: for Stas, because of an unsuspected impurity in his silver had obtained less silver chloride from a specimen of the metal than really should have been produced by it. This led him to an atomic weight of chlorine distinctly too small and to values for sodium and silver distinctly too large. The excessive value for silver was augmented still further by the fact that his method of precipitating the silver chloride by placing fused common salt in a solution of silver nitrate tended to cause impurity in the precipitate. But these are technical details already beyond the compass of this brief address. The higher atomic weight of chlorine has been amply justified by others in many ways, and there can be no question to-day of the trustworthiness of the outcome.

Not only were Stas’s values found to be perceptibly in error, but the reasons for the errors were made clear. An easily understood fault of Stas’s work was the use of such very large quantities of material that it could not be properly purified. There is no use in weighing a given specimen of silver, for example, to within one part in a million when its undetermined impurities amount to fifty parts in a million.

Following this research, a number of others on other alkali metals, on sulphur, and on nitrogen were carried out, because these elements were more or less involved in the revised ratios just mentioned. New values for all these elements were found, more or less divergent from those previously accepted.

The next important contribution, involving a new method and a somewhat new problem, was that in which lithium and silver were compared directly with oxygen through the analysis of lithium perchlorate and the precipitation of the chlorine in lithium chloride by silver. These two processes together give a new means of comparing oxygen with silver – an eventuality much to be desired, because many atomic weights are determined by reference to silver, and all are stated in relation to oxygen. Here again the methods of preparing pure lithium perchlorate, of proving it to be free from water and to have remained wholly undecomposed in the process of drying, and of analyzing it without loss, are all too complex for detailed discussion. The comparison of lithium chloride with silver was simple enough, resembling precisely the case of sodium. Silver was found to have an atomic weight at least as much smaller than Stas’s, as previous estimates of the impurities in Stas’s silver had indicated. On the other hand, all of the errors having been heaped by Stas upon the head of the lightest of all metals, lithium, they led to an excessive estimate of its atomic weight by a whole per cent – apparently his largest error.

Years ago my present colleague, Gregory P. Baxter, and I had worked upon the atomic weight of terrestrial iron, and found it to have a value distinctly lower than that usually assigned to it. In connection with neverceasing curiosity as to the constancy of the atomic weights, I wondered later whether or not iron in meteorites, possibly having its birth far beyond the limits of the solar system, might have a different atomic weight from ordinary iron. Baxter kindly consented to investigate this question and, with characteristic care, using methods similar to those used upon terrestrial iron, he found that the iron from meteorites has precisely the same atomic weight as this metal smelted on earth. The outcome, although not unexpected, is, nevertheless, of interest, and thrills one who appreciates it with an added realization of the unity of the Universe.

Investigations on calcium, carbon and sulphur, which verified the earlier work of others, may be passed over without comment ; but the most striking, perhaps, and most puzzling of all the results of the work of thirty years, namely, the amazing situation with regard to the metal lead, deserves consideration in detail.

If, as Boltwood suggested, metallic lead is the solid end-product of radioactive disintegration, the lead found in radium minerals acquires a new and extraordinary interest. It must then be an elementary substance which has been produced on earth in geologic times. The question as to whether or not its atomic weight is equal to that of ordinary lead of far more ancient origin is a question of no common importance.

Through the kindness of several workers in radioactivity, I had the privilege of being one of the first to compare the atomic weights of radioactive lead and common lead. Dr. Fajans, Sir William Ramsay, Professors Boltwood and Ellen Gleditsch, Mr. Miner and others, at about the same time, were so good as to send to Harvard specimens of radioactive lead for examination. With the help of several assistants, I was able to show that the pure lead from crystalline radioactive minerals has an atomic weight as low as 206.08, instead of the 207.2 of ordinary lead. This result has been confirmed by the more or less simultaneous investigations of Madame Curie and Otto Hönigschmid. The study of other properties of the metal and its salts shows that this difference of weight is the only important difference between the two kinds of lead. Their atomic volume, electrical behavior, etc. are identical, and their spectra are very nearly, if not exactly, alike. Ordinary lead was shown by Baxter to possess a constant atomic weight, no matter what its source, provided that uranium minerals were not present.

The existence of two kinds of atoms of lead is of great theoretical interest, giving us new ideas as to the ultimate nature of the elements. It may also give us new geophysical arguments in constructing a new cosmogony – since it affords an unexpected clue as to the history of the various accumulations of material on the earth. No one knows how many other elements may also possess so-called isotopic forms of this kind. Of course even if this should often be the case, the actual values already found are the important ones for practical purposes, since they represent the actual substances found on earth. For theoretical purposes, on the other hand, not only must we learn how to separate the isotopes of lead (which are so exactly like that no ordinary chemical methods will suffice); we must study likewise all other elementary substances, in order to find out whether they also may have atoms of differing weight. This quest has already been begun by several experimenters. Who knows what modifications our Periodic System of the elements may suffer, and what illumination it may gain, from such experiments? Already the explanation, which has been offered by several chemists independently, concerning the reason for the existence of the isotopes of lead, is highly suggestive.

The discovery of radioactivity with its amazing concomitants has thus put quite a new aspect upon the study of atomic weights. Because of this recent knowledge, some investigators seem to believe that these so-called “constants of nature” are of less significance than of old, since some atoms may be transitory, and since a given substance may have different atomic weights according to the circumstances of its life history. Nevertheless, when this new theory of elemental relations was at a crucial stage in its development, eager investigators all over the world appealed to the student of atomic weights for definitive decision of their problem. Such a widespread appeal does not indicate a lessened significance in atomic weights, but rather a new significance. The atomic weight of a given specimen of elementary substance no longer necessarily gives certain evidence as to the atomic weight of a specimen of fundamentally different origin. But on the other hand, the several values of the atomic weights are not only characteristic of the specimens concerned, but also give us perhaps the most certain clue as to their origin and history. This is indeed a new significance. Thirteen years ago, before the possibility of radioactive isotopes had been suggested, I took occasion to point out the wider bearings of such a variation, if it should ever be found:

“The question as to whether or not the supposed constants of physical chemistry are really not constants, but are variable within small limits, is of profound interest and of vital importance to the science of chemistry and to natural philosophy in general. If this latter alternative is true, the circumstances accompanying each possible variation must be determined with the utmost precision in order to detect the ultimate reason for its existence. As Democritus said long ago, “The word chance is only an expression of human ignorance.” Every variation must have a cause, and that cause must be one of profound effect throughout the physical universe. Thus the idea that the supposed constants may possibly be variable, adds to the interest which one may reasonably take in their accurate determination, and enlarges the possible field of investigation instead of contracting it.”

The subject of atomic weights is thus far from being a completed and closed chapter of Science. The future opens up a prospect of almost endless further investigations, because the study of a single kind of material is not enough to make sure of the universality of an atomic weight. The work which has been described to-day is thus only a beginning.

Each generation builds upon the results of its predecessors. Stas improved the admirable work of the master Berzelius, using many of his methods, with improved appliances and wider chemical knowledge of the later date. As you have just seen, the more recent researches have improved upon those of Stas. In years to come, let us hope that yet finer means of research and yet deeper chemical knowledge may make possible further improvements, yielding for mankind a more profound and far-reaching knowledge of the secrets of the wonderful Universe in which our lot is cast.