but that it must be made more fusible by mixing with it a little double chloride of sodium and aluminium and with potassium chloride; and it is the proposition of De Ville to use cryolite in this way as the main part of the bath and an anode of carbon and alumina in accordance with the foot-note, which the defendant's counsel and experts contend is substantially the same process as that on which Hall claims a patent. Cryolite, as we have seen, is a double fluoride of aluminium and sodium. The theory of the process of De Ville is that the electric current disrupts the fluoride of aluminium, which is a less stable compound than the fluoride of sodium. Fluorine gas is liberated at the anode, and, being presented to the mixture of carbon and alumina, unites with the aluminium and leaves the oxygen with the carbon to make carbonicoxide gas. The fluoride of aluminium thus constituted at the anode renews the bath and is supposed to keep it constant, while the aluminium of the original fluoride of aluminium which was in the bath is deposited at the negative pole.

We are very clear in our opinion that this is not an anticipation of Hall's patent. The primary and most important step of Hall's patent is the dissolving of the alumina in the fused bath of the double fluoride of aluminium and sodium. This happens before any electrolysis occurs at all. When the electric current is applied it disrupts the alumina, sending the aluminium to the cathode and the oxygen to the anode, where it unites with the carbon anode and becomes carbonic-oxide gas. The bath material in which the alumina was dissolved—that is, the double fluoride of aluminium and sodium-is not affected by the electric current, and the process is made continuous by the simple addition of alumina to the bath. In the De Ville process, however, aluminium is not dissolved in the bath. The fused mass subjected to the current is composed of a double fluoride of aluminium and sodium, with no alumina in it. No effect comes from the presence of the alumina in the anode until after the electric current is passed through and the fluoride of aluminium has been disrupted and fluorine gas has been produced at the anode. Then the gas unites with the aluminium of the alumina and restores the bath, in a manner, to its former state, to be again disrupted by the application of the current. In the one case there is a simple solution of the alumina in a bath and its disruption by the current, the bath remaining constant, while in the De Ville process the bath itself is disrupted and restored again by the union of one of the disrupted elements with an element of the alumina in the compact anode.

In the Hall process there is simply solution of the alumina and electrolysis. In the De Ville process there was electrolysis followed by what is called an electrochemical solution of the alumina.

Defendant's counsel and experts, in their effort to make the two processes the same, rely upon the supposition that when the compact anode of carbon and alumina was inserted by De Ville into the bath of cryolite 10693 PAT-37

some of the alumina was necessarily dissolved in the cryolite, and so the current electrolyzed that alumina, as in the Hall process. In the first place, the experiments made on behalf of the complainant demonstrate that a compact mixture of carbon and alumina (which can only be made by the use of some binding material) may be kept in a bath of cryolite for many hours without dissolving the alumina at all. Some counter experiments by defendant are by no means so satisfactory, because the results may be attributed to the presence of other causes than the solution of the compact anode.

But suppose it to be a fact that in De Ville's process alumina was dissolved in the bath from the anode, and that thereupon it was electrolyzed, as in the Hall process. It was a mere accident, of which De Ville made no note, and which, therefore, we may reasonably infer he did not observe. Accidents of this character cannot be relied on as anticipations of a patented process (Tilghman v. Proctor, C. D., 1881, 163, 19 O. G., 859; 102 U. S., 707, 711) when the operator does not recognize the means by which the accidental result is accomplished and does not thereafter consciously and purposely adopt such means as a process for reaching the result. De Ville did not intend that the alumina should be dissolved in the bath. If he did, why did he unite it with carbon by means of pressure and a binding material? If he had intended to dissolve the alumina, he could not have intended anything which would have so interfered with his purpose. Just as Professor Langley says, such a course would have been as reasonable as it would be for a man wishing to dissolve a lump of sugar in his coffee to varnish it or grease it. De Ville did not intend to dissolve the alumina in the bath, first, because, as we shall see hereafter, he had no knowledge that it would dissolve therein, and, second, because until the fluoride of aluminium had been decomposed in the bath by the current he did not need anything with which to restore that fluoride to the bath. He felt that it was necessary to have the alumina in a solid form in close connection with the carbon, because without the strong attraction that the carbon would have for the oxygen of the alumina the fluorine gas could not get at the aluminium of the oxide.

De Ville was seeking to electrolyze the fluoride of aluminium, while Hall was seeking to electrolyze alumina. It was a condition precedent to Hall's success that his alumina should be in a liquid state. As we have seen, it was essential to De Ville's process that this alumina should be where it was-i. e., mixed in solid form with the carbon.

This brings us to a theory of the Пall process which has been advanced by one or more of the defendant's experts to show that the Hall and De Ville processes are alike. Hall says in effect in the specifications for his patent that the electrolyte in his process is the alumina held in solution and that in the decomposition the aluminium of the alumina goes to the cathode and the oxygen of the alumina goes to the anode, the fused double fluoride constituting the bath remaining unaf

fected. The theory-or, we would better say, the alternative theoryof the defendant's experts (for they venture more than one) is that the fluoride of aluminium is the electrolyte, and that when it is decomposed the fluorine attacks the dissolved alumina, drives off the oxygen to the anode, and unites with the aluminium remaining, thus restoring or regenerating the bath. This theory is advanced for the purpose of showing that the Hall bath does not remain constant as he claims, but has to be regenerated, as in the De Ville process, the only difference being that in the former the alumina is held in the bath in solution, while in the latter it is held in a compact mixture of carbon and alumina. If this were the only difference between the two processes, we should hold that the Hall process was a patentable discovery, because the regeneration of the bath from the dissolved alumina, if such were the chemical action which took place, would be a successful process, because regeneration is so complete as not even to betray itself, while in the practical operation of the De Ville process the regeneration of the bath is very incomplete. The fluorine gas escapes into the air and is so corrosive as to injure both workmen and apparatus, and the process is a failure for commercial purposes. The compact mixture of carbon and alumina is hard to make and obstructs regeneration, and the substitution for it of dissolved alumina, even if their functions in the process are exactly the same, is the step fron failure to success.

But there is no satisfactory evidence at all that the dissolved alumina is not the electrolyte, and the burden is on the defendant to show it. (Coffin v. Ogden, 5 O. G., 270; 18 Wall., 120, 124.) Professor Chandler, of the Columbia School of Mines, called as an expert witness for the complainant, demonstrates, so far as chemical actions and reactions are capable of demonstration, that it is the alumina which is decomposed by the current. He says that the alumina dissolves in the fused bath of cryolite as sugar does in water, quietly, without any of the usual evidence of chemical action, such as the production of heat or the reduction of fusibility or thickening of the liquid. This seems to show that the alumina in solution remains chemically unchanged. He refers to the fact, admitted on all hands, that when the current is applied aluminium goes to the cathode and oxygen to the anode without any indication of any other product of decomposition. This result is fully satisfied by the explanation that the alumina alone is decomposed. He refers to another fact, also admitted, that as soon as the alumina in the bath is exhausted the resistance to the current and the voltage to the current to overcome it become doubled, and are not reduced until alumina is added to the bath. This would seem to show that the electrolyte has been changed by the exhaustion of the alumina, and therefore that, when present, alumina is the electrolyte.

The professor further shows from the observations of the most reliable chemists that the heats of formation of the three compounds in the

bath are such that it would require four and eight-tenths volts to decompose sodium fluoride, four volts to decompose aluminium fluoride, and two and eight-tenths volts to decompose alumina, which proves that the alumina would first yield to the action of the current and in the presence of the other two compounds would be the electrolyte.

More than this, Professor Chandler points out a fatal contradiction in the theory that the fluoride of aluminium is the electrolyte in the Hall bath. That theory assumes, on the one hand, that the dissolved alumina will be decomposed by the fluoride of aluminium, and, on the other, that the alumina is a more stable compound than the fluoride when electrolysis ensues, so that the fluoride and not the alumina becomes the electrolyte. Finally, the theory is shown to be unfounded by the total absence at the anode of fluorine gas, which would certainly be there present in perceptible quantities if fluoride of aluminium is the electrolyte and fluorine is released by the current.

No facts of any kind are advanced by the defendant's experts to meet this demonstration of Professor Chandler, which is fully concurred in by Professor Langley and Doctor Raymond. The arguments, such as they are, depend on analogies drawn from the electrolysis of entirely different compounds, as, for instance, water. With the burden on the defendant on this issue, we are very clear that it has not been sustained. An alternative theory advanced by some of defendant's experts is that when alumina is dissolved in the bath a compound of oxygen, fluorine, and aluminium, called oxyfluoride of aluminium, is formed, which is the electrolyte. It is sufficient to say of this theory that no chemist, and certainly none of defeudant's experts, ever isolated such a compound, and no one knows that it exists. The reason for its theoretical existence is that such a compound is formed with the chlorides. Analogies between the fluorides and the chlorides constantly fail. In this very case we find that the former will dissolve alumina and the latter will not. The oxyfluoride of aluminium theory is mere specu lation and not worthy of consideration in a case involving property rights.

The defendant company instituted a number of experiments to show that the De Ville process would work as well and in much the same way as the Hall process. The evidence concerning these experiments, although it all comes from the defendant's witnesses, leaves no doubt in our minds that they were not conducted in a fair spirit, but rather with an intense desire to force a demonstration of the identity of the De Ville and the Hall processes. Nothing else can explain the palpable disobedience to De Ville's express directions for carrying out his proc ess. According to De Ville's directions the bath must not be heated up to the melting-point of aluminium. This is evident, because his main process contemplates an aluminium anode, and heat beyond its melting point would destroy it as an anode, and it is for this reason that he directs that the fusibility of cryolite be reduced by mixing it

with chlorides. In the experiments of the defendant the bath was heated far beyond the melting point of aluminium. It was so hot that the carbon of the carbon-alumina anode, a large part of which was not sunk in the bath, but was exposed to the air, united with the oxygen of the air and was burned out. This left on the exterior of the anode nothing but a ring of pure alumina, which, dropping into the bath, was of course dissolved, as in the Hall process, and from it aluminium was deposited at the cathode. Then, too, stubs of the partly-used carbonalumina anodes were ground up and thrown into the bath of cryolite, and of course the particles of alumina separated by the grinding process from the carbon became dissolved in the bath, and the Hall process was again reproduced. De Ville did not contemplate the presence of alumina in the bath except in a compact mixture with carbon, and yet the experiments of defendant presented alumina to the bath wholly free from carbon. Even with those approaches to the Hall process the cost of producing the aluminium which was produced exceeded that of the Hall process some sixfold. Nothing could more clearly demonstrate the difference between the De Ville and the Hall processes.

We have considered the Hall process as a whole, and thus compared it with the De Ville bath-regenerating process. A comparison of his process with De Ville's and the difference would not be complete, however, without some reference to Hall's claim that he was the first discoverer of the fact that alumina would freely dissolve in the fused double fluoride of aluminium and sodium. The fact is that such a bath will dissolve from ten to twenty-five per cent. of its weight as completely as water will sugar. Defendant claims that this was well known in the art long before Hall's application. The first publication relied on to establish the claim is found in one of De Ville's works, as follows:

Cryolite is a double fluoride of aluminium and sodium which one may produce artificially or imitate by mixing hydrofluoric acid in excess in pure and calcined alumina and carbonate of sodium in quantities such that the sodium and aluminium will be present in the proportions in which they exist in cryolite. After drying and fusing the mixture, one has a limpid and homogeneous substance which possesses all the exterior characteristics of fused cryolite. I have not yet made an analysis of it, but the weight of the substance produced is such that one must suppose that the alumina and soda have lost all their oxygen by transformation into fluoride. This cryolite, like the natural cryolite, gives aluminium when reduced by sodium. It also yields aluminium under the influence of an electric current, which a mixture of alumina and fluoride of sodium fused together will not do. When this experiment is made, one perceives that the alumina dissolves in the fluoride, but in small quantity, and remains in the condition of alumina because an electric current which passed through the well-fused substance yields sodium and fluorine. These experiments, which succeed very well when one employs a mixture of fluoride of sodium and fluoride of potassium, proves further that alumina at red heat is decomposed neither by sodium nor by potassium.

There is no statement here at all that cryolite will dissolve alumina. The words "when this experiment is made," etc., clearly refer to the experiment last spoken of-that is, to passing an electric current