The court below dismissed the infringement suit brought by the appellant, as plaintiff, based upon the Bentson patent, No. 1,246,527, dated November 13, 1917, for an improvement in trenching machines. This patent had to do solely with an improvement, consisting of a spring cleaning device, by which the buckets on the continuous chain would be automatically cleaned by a resilient scraper. The court found that the patent involved invention and was infringed, and those points are not discussed in this court; but the patent was found invalid because of sale or public use by the inventor more than two years before the application. This two-year period began January 10, 1911.

The general legal principles involved are well settled. If the critical public use is by or under the control of the inventor, and for no longer period than is reasonably necessary to determine by experiment whether the invention is complete, or requires modification or change before final adoption, or if the sale, though otherwise complete, reserves to the inventor a similar right of experimentation and substitution, then, although there technically may be public 'use or sale, that which has happened is within the implied reservation in the statute, and does not invalidate the patent. Smith & Griggs v. Sprague, 123 U. S. 249, 256, 8 S. Ct. 122, 31 L. Ed. 141; Jenner v. Bowen (C. C. A. 6) 139 F. 556, 561, 71 C. C. A. 540. The ■ proper extent of such experiment will depend upon the subject-matter. Since the durability of a pavement can only be determined by long use; the maximum period of test and experiment may be had (Elizabeth v. Pavement Co., 97 U. S. 126, 24 L. Ed. 1000); while, if every question involved will be answered by a few minutes’ or a few hours’ test, then the allowable period is reduced to the minimum (Jenner v. Bowen, supra).

Properly to apply these rules to the unusual and probably unique facts of this ease requires a careful statement. It is mainly in the inferences to be drawn from the facts, rather than as to the facts themselves, or as to any legal rules, that we find ourselves unable to agree with the trial court.

The inventor, Bentson, was superintendent of plaintiff’s predecessor, having charge of its factory near Chicago, where it was manufacturing a line of machinery, including these ditch-digging or trenching machines. During the summer of 1910 it produced a smaller model intended for ditches in farm draining. In September, 1910, it had not only developed and completed some machines of this model, but had improvements, by way of changes or attachments, which were in process of development, and which it intended, if successful, to attach to this and other models. It was desirable to test these improvements, not only on the company’s testing ground at the factory, but under the varied conditions that would arise in the field, with different soils and unknown difficulties. Anderson had work of this kind available not far from the factory, and wanted to buy a machine, but had no means with which to make the requisite down payment; Bentson saw the opportunity for extending his testing field, and proposed to Anderson an arrangement which was, as between them, in substance a partnership. He furnished to Anderson the money for the down payment. Anderson bought the machine, taking the bill of sale thereof, running to him, and giving a chattel mortgage to the company for the deferred payments, and it was arranged that the operating profits should be used, first, to pay Anderson a daily wage; second, to pay the unpaid purchase price; third, to repay to Bentson his advance, with interest; and, fourth, the remainder to be equally divided between the two.

The machine so sold and delivered did not contain the improvement now in question, or some others that were more or less definitely in mind. That whole subject was covered by an oral understanding by Anderson with Bentson, who was acting for himself and for the factory. As somewhat vaguely stated by Bentson and Anderson, it was arranged that the machine was to have all the improvements made in 1910; in exchange for this. Bentson and the company had the right to subject the machine to use for their experiments in determining what were and what were not desirable improvements. In return *699for the present help received from Bentson, and the indefinite expectation of additional improvements and attachments, Anderson •was willing to snbjeet his free nse of the machine to this burden. Bentson was willing to risk his money in furtherance of whatever personal benefit he might indirectly get from perfecting the contemplated improvements, and the company was willing to let Anderson have the additional improvements, if and when perfected, in exchange for his co-operation in permitting the use of his machine for field experiments.

The machine was delivered and went to work a few miles from the factory about mid-September, 1910, and continued at work in this vicinity more or loss steadily for a period of six or seven weeks. During this period Bentson was very frequently observing its work and conducting experiments upon it. During this period three different attachments or improvements were added, and after trials and*changes were finally abandoned.1 Perhaps as the result of watching this work, or perhaps from some previous idea, Bentson planned out the spring scraper in question. A single device was built with parts forged in the blacksmith shop, and probably in September was put in the • machine. From time to time several changes wore made in this device — not, it is true, in matters affecting the theory of its operation, but in matters of shape, attachment, and strength, necessary to determine whether the invention was practical and useful, so as to he of commercial value. It was finally decided that the device in its existing form was not satisfactory, because with the 12 or 16 inch buckets with which the machine had been provided there was not space enough for the seraper to operate satisfactorily in those very sticky soils where it was most needed. It was believed that it would be more satisfactory with larger buckets.

Accordingly the machine was taken into the factory, remaining there while it was considerably overhauled; 24-inch buckets were installed and the scraper attachment— still with the hand-forged parts — was adjusted and modified for the altered machine. It was then capable of different work, and some charges for the change were made against Anderson. He then took it to its new location, about 50 miles away, where it went to work with different soil and conditions. This second period lasted about four weeks, until the ground became too much frozen. During this second period Bentson attended and observed the work several times. One other attachment was tried out and rejected; and this use was experimental, and under Bentson’s control, both as inventor and as joint owner, not less in theory, though somewhat less in practice, than during the former period. As the result of this second period it seems to have been concluded that the scraper attachment, though not sure to give satisfaction at all times, probably could not be further perfected, and it was allowed to remain on the machine, which was then laid up until spring. At some unfixed time in January or February, 1911, the factory order was given, and the drawings made for eastings to he used in the first lot of machines to be regularly provided with the spring cleaners.

To onr minds these facts make a clear ease of that merely experimental use by or for the inventor, which is not sufficient to be the statutory public use to defeat the patent. They make the situation clearly distinguishable from any decision urged in support of the result below. This case is classifiable with the Elizabeth rather than with the Jenner precedent. These machines a,re of great size, np to 50 tons; they operate often far from a repair point; they must not break down, in any detail, under extreme burden and strain; they must dig anything from mud' to hard-pan; this model never did succeed commercially, and was abandoned. The undisputed evidence is that one entire season’s use, 10 months, was necessary to determine success or failure; and the court should not fix a shorter period, unless the conduct of the parties is unequivocal to the contrary.

. So we come to the matter of being “on sale,” as distinguished from public use. The sale of the machine to Anderson was by the paper writings complete and unconditional; but this attachment was not then in existence. Anderson got no claim to it, save by the oral promises of the vendor and the inventor that he should have any improvements, and the same oral contract made the machine subject to experimental use by the inventor, to develop and test this or other inventions — -all without Anderson’s further consent, or even against his protest. Inquiry just when, from the viewpoint of Anderson’s creditors or an insurance company, the title to the material in this attachment vested in Anderson, is not helpful. It seems clear enough that, even if it *700did for these purposes belong to Anderson as soon as it’was put on, his right as between the parties remained conditional, and subject to the manufacturer’s right to take it off and substitute something better for the reputation of the machine. '

_ The sharpest test on this matter of sale would come if we might assume that at the end of the second use, in mid-December, the inventor and the maker had definitely con-eluded that the spring scraper test was ended, that the device should remain on. the maehine as it, was, and that Anderson could use it without further interference or control by the others. Upon that supposition it might be difficult to escape the conclusion that there was then a complete and perfected sale of the entire machine, including this attachment. The record does not justify that posirxr . , t . tive assumption. We are -not able to say ... .A ,, . * mi-, .. _ that, even m the spring of 1911, it would not have been Anderson’s duty to permit this attaehment to be removed. He was entitled only to the improvements; nothing was an improvement which would not work reasonably well; the maker’s judgment on this subjeet would be important, if not controlling; and there is no proof that its judgment had before January 10, 1911, been so expressed as to be irrevocable.

The presumption of the validity of the patent is such that the defense of invention by another must be established by the clearest proof — perhaps beyond reasonable doubt. The same rule apparently should apply to the defense of prior public use or sale by the inventor. When an actual sale in the critical period appears, it may well be that the trier of fact will mentally shift the burden of evidence necessary to show this sale to have been so modified that its existence did not make the device on sale” within the meaning of the statute; but we see no reason why the legal burden of proof should shift, and we know of no authoritative and considered decision to that effect. It would seem that the legal and heavy burden of proof as to all the elements involved continues until the end upon one who attacks the patent grant.

So far we have in the main considered the sufficiency of the sale for the purpose now involved as if it had been to Anderson alone, Our conclusion is confirmed by the inventor’s position as joint vendee. If an inventor gets the manufacturers to make a machine for him, .with which he may experiment and test his invention, the transaction clearly does not make the invention “on sale” at the moment of delivery of the machine, even though, as between the parties, it is a sale. The principl0- d0es not seem to be different if the inventor takes in a partner in the ownership of the machine, and that is what happened in this case.

The trial court regarded the Anderson use and sale as the strongest and best proved of g^y 0£ £hQSe charged to have occurred before January 10, 1911. We .agree in this eonelusjon, and we do not discuss the others. The decree is reversed, and the ease remanded for appropriate further proceedings,

The defenses of anticipation, noninvention, and nonfringement were open to appepee on this appeal, as reasons why the deore0 be;ow -^as right. They have not been ¿resented_ We have assumed that appellee did ^ ^ ^ ^ t]lem. in the in-the end iitigation, our finding , . . that the decree below was erroneous is na- , _ _ , T«. tended to conclude all these questions. If . , , , ^ . ^e are wrong m our stated assumption, we will. entertain an application for rehearing on these Points.

^, DONAHUE, Circuit- Judge, dissents.

On Application for Rehearing,

1. The statement in the opinion, that “the legal and heavy burden of proof as to all the elements involved eoiddlllles until the end upon one who attaoks pateQt grant„ we made deliberate. ^ though' familiar with Smith v. Sprague, 123 U. S. 249, 264, 8 S. Ct. 122, 31 L. Ed. 141; Columbus Co. v. Standard Co. (C. C. A. 6) 148 F. 622 629, 78 C. C. A. 394, and others of t]ie eases now mged. Some of t]lem contain at least dieta to fte effeet tkat the burden may ski£ t> bllt we ar6 not satisfied that the criticized statement ^ our opinion is incorrect; the relative unimportance of the matter in this ease does not justify a complete review and analysis.

2. The application suggests an ineonsistency in saying, as the opinion did, that in Deeember it seemed that the invention probably' could not be further perfected, and yet holding that the experimental period did not expire until January. This suggestion overlooks the underlying inference of the opinion that, when the ditcher was laid up in Deeember for the winter, it had not been decided whether this particular attachment would ever be manufactured, or would be thrown away as an unsuccessful experiment. Not until the order for castings was given in January did this uncertainty disappear,

3. The remaining matters alleged were sufficiently considered in the opinion,

The application is denied.