Since the beginning of my chemical pursuits, I have been attentive to the philosophical questions involved in the classification of the elements. Interesting relationships had already been pointed out by several scientists (especially by Dumas) between the atomic weights of certain simple substances and the properties, which invited the chemists to form a natural family of these substances.

I endeavored find new common features by comparing either the atomic weights elements or their qualities, such as, for example, the emission of luminous rays of definite wave lengths at high temperature. Thus I arrived at still unknown relationships, and I drew from them some deductions that seemed quite intriguing to me.

Among the conclusions to be drawn from my attempts at chemical classifications was the probability that unknown elements existed that could fill the vacant places in the natural series.

It is clear that the position of a hypothetical substance in a natural family approximately indicates the properties of this substance.

The uncertainty that exists concerning the exact chemical reactions of a hypothetical substance that is defined only by its position in a natural series renders it rather problematic whether success can be achieved by relying exclusively on such precalculated reactions, because the slightest error in predicting one of these reactions could throw the sought substance out of the place in analysis assigned to it by the theory.

The difficulty seemed very great to me. To overcome it, I devised a particular route of mineral analysis, such that an error concerning the properties of the sought substance or even of known elements would not prevent a final success.

By its extreme sensitivity, spectral analysis is of great help in this kind of work; nevertheless it is not an essential and indispensable part of my method of research. [Spectral analysis] is a marvelous tool, and I have devoted long years to making it more perfect. Before the modifications I developed for the equipment used to produce electrical spectra, it would have been impossible to examine spectroscopically the small quantities of liquid in which I established the existence of gallium for the first time.

My first attempt at a search for new elements started fifteen years ago. At that time I had no laboratory, and my instruments were totally insufficient. That attempt, carried out with considerable quantities of material, had to be abandoned; most of its products were lost. In 1863, my present laboratory was constructed and much better equipped. I renewed my attempts and made a series of studies, but without success. Evidently I bad used too little material.

Finally, I decided to operate on a larger scale, as I had done in the beginning of my work, and in February of 1874 1 started to treat 52 kg. of blende from Pierrefitte, acquired for this purpose in the summer of 1868.

On August 27, 1875, between three and four at night, I perceived the first indications of the existence of a new element that I named gallium in honor of France (Gallia).

Guided by certain considerations, I admittedly did not wait for the rigorous progress of my methodical analysis. Thus, on August 27, I took some of the white precipitate that started to form in one of my products on contact with zinc foil. This precipitate was dissolved in hydrochloric acid, and the solution was precipitated with excess ammonia, filtered, evaporated, and the ammoniacal salts destroyed by boiling with aqua regia. When the solution thus obtained was submitted to the action of a spark, the spectroscope showed numerous known lines and, besides, the very weak trace of a violet line at 417.0 on the scale of wave lengths.

This line did not exist in any of my pictures of spectra. I did not doubt that I was involved with a new element, and I immediately applied myself to increase my supply of the precious material.

I estimate that the quantity of gallium contained in the small drop examined at the time of my first observation did not exceed 1/100 mg.

All of the white precipitate that had formed was, therefore, dissolved in hydrochloric acid and treated as described above. The resulting acidic liquid was saturated with hydrogen sulphide, filtered, and again treated with hydrogen sulphide after adding an excess of ammonium acetate. A zinc sulphide separated out which, dissolved in hydrochloric acid, clearly gave the 417.0 line in the spectroscope and, besides, another, weaker violet line near 403.1. Later observations proved that the 403.1 line also belonged to gallium.

After three weeks, I finally accumulated 2 to 3 mg. of gallium chloride, still mixed with zinc chloride.

I then went to Paris where I had the honor, in the last week of September, 1875, to carry out a series of experiments to demonstrate the individuality of gallium in the laboratory of Wurtz before the section of chemistry of the Institute.

Thanks to the kindness of the societies of Vieille-Montagne, Nouvelle-Montagne, and Corphalie, of my learned friend Friedel, and especially of Malgor, the engineer who directs the operation of Pierrefitte for the Société l'Asturienne, I assembled a considerable quantity of minerals, which were treated directly for gallium.

Gallium was reduced to the metallic state for the first time in November, 1875, by electrolysis of an ammoniacal solution of its sulphate.

I have recently put together and treated all the gallium-containing products in my possession; have extracted from it 0.65 g. of pure gallium. Such is the yield from about 435 kg. of raw materials.

I finally united the six samples of pure gallium mentioned before into one quite homogeneous piece. Its specific gravity at 23° (relative to water at 23°) was:

The prevision of Mendeleev is thus exactly verified.

Among the several hypothetical elements indicated by the ingenious classification of Mendeleev, there is one that seems to refer to gallium in the calculated properties. . . .

I must say that I was unaware of the description given by Mendeleev for his hypothetical element. I should even add that my ignorance has perhaps been favorable to me, in the sense that I would have experienced delays had I been led to search for gallium in the precipitates formed by ammonia and not in the ammoniacal solutions, which retain it completely, or almost completely, when it is present in small quantity.

The low melting point of gallium also seems difficult to reconcile with what the theory permitted to predict.

Thus, in spite of the incontestable merit of Mendeleev's hypothesis, several reactions and qualities of the new metal differ sufficiently from what the theory indicates to have made it quite problematic whether research guided only by that theory would have been successful when extended to a mineral that is very poor in gallium.

Therefore, it seems probable to me that neither the calculations of Mendeleev nor my own hypotheses would have led to the knowledge of gallium for a long time without the particular experimental method I have followed.

However, the discovery of a new metal now invests the classifications, enabling us to foresee the existence of unknown elements, with an importance they would not have been accorded if no positive fact had come up to support the ideas. I need not emphasize the great interest connected particularly with the confirmation of Mendeleev's views concerning the specific gravity of gallium.