L. HAND, Circuit Judge

(after stating the facts as above). On the' main issue we think that the towns are right; the contractor confessedly did not dig to the solid ledge. That was a breach, and it was the cause of the bridge’s collapse. We did not hold before, and could not properly have held, that the contractor might ignore this promise; but wo did say that the specifications and the plans, when read together, assured him that what seemed like solid ledge at tho prescribed depths he might accept as such. Thus, although his engagement remained to dig to the ledge, and though he was in default, ho woftld have had an excuse, had he gone to the lines shown and found what ho supposed to be the ledge. Yet we cannot see what use he can make of this representation, if he did not go to the elevations shown. There was no possible warrant in the contract for supposing that what he might find at lesser depths was the ledge he was seeking. It might be, or it might not, and he must take his chances. If he failed to go to the agreed line, he had not relied upon the representation of the contract.

Nor can we accept the contractor’s argument that the specifications and plans should be construed as moaning no more than that he should dig to the solid river bed. The engineers apparently did believe, not only that the lines approximately coincided with the river bed, but also that the ledge formed that bed. Perhaps it was natural for the contractor to conclude that tho bed alone was intended, and to stop there; but that was not what he undertook when he agreed that the "excavation '* s' * shall be done to the line shown.” That line meant the elevations shown on sheet two of scheme B, and these be neither reached nor tried to reach.

We agree that nobody could expect to find a solid ledge of straight horizontal surface, and we cannot, therefore, accept the dilemma, which the towns put, that, if the contractor did not dig to the line, he had not relied on the contract, while, if he went below it at any place, ho knew that it was wrong. But they do not need this dilemma to succeed. It is enough if the plans meant, as we think they did, that whatever seemed solid a.t or below the lines was ledge. Perhaps, indeed, there was room for trifling discrepancies.. Variations of a foot or less possibly were permissible, but in the case of the Vermont pier, which is the one that collapsed, the excavation was always between 2 and 6 feet above the line, and this in a foundation itself only 18% feet in depth. No latitude of construction can tolerate such deviations as that. Thus there was a substantia], and indeed a very serious, failure to comply with the contract.

The breach stood therefore bare, and to escape the contractor had to show some other excuse. He tried to do so by Marshall’s evidence. We may assume, without deciding, that the reasonable implication of Marshall’s conduct was that he supposed the cofferdams to have been sunk to the prescribed levels. True, he did not say so expressly, but it was at least permissible, considering Marshall’s apparent activities, to conclude that he would have objected, had he thought the work was not being properly done. We need not, how*258ever, decide whether, if Marshall had been authorized to speak for the towns, this would have made out an excuse for performance, or have been a practical construction of the contract. We may assume as much arguendo, because the specifications, Under the heading “Consulting Engineer,” in any ease dispose of the contention. The language is: “The assistants’ approval of improper work or material shall not entitle the contractor to payment therefor, if the same shall not be approved by the board.” The board never knew that the contractor had not dug to the elevations shown; they knew nothing whatever about the matter. Storrs, it is true, swore that he knew that the piers were approximately down to the line, in which he was quite mistaken. Perhaps he thought so because Marshall had told him; but his approval, if he gave it, would not help the contractor. The clause meant that no approval should count, unless the board, being advised that the work was improper, was content to accept it. Only so could it be protected from the negligence or disloyalty of the engineer’s assistants. Nobody suggests that any .such consent was 'ever given.

Next, as to the examination of the bottom by Storrs personally: After the contractor had sunk the cofferdams, and by a diver had examined and somewhat ineffectually cleaned the bottom, and when the time had therefore come when it proposed to begin to pour the concrete, Storrs was summoned to examine the foundation, and came to the site. His story, which the jury accepted, was that he pushed about with an iron rod and found rock and sand. The contractor’s foreman on the job was present, and Storrs asked him whether the rock he felt was the ledge. The foreman said that it was, to which Storrs replied that, if this was true, he might begin to pour. Nothing more was said, though Storrs admitted that in fact he felt sure that the ledge had been reached.

The contract required the excavation, not only to go “to the line shown,” but also to be done “to the satisfaction of the engineer.” These are separate stipulations in the copula, and, unless the engineer had power to excuse the first, his satisfaction with the excavation was not enough. The contractor insists that he had, but we cannot agree. While we concede that it is not necessary that a contract shall in so many words declare that the decision of an engineer shall be “final and conclusive” (U. S. v. Hurley, 182 F. 776 [C. C. A. 8]), nevertheless the equivalent must appear within its four corners (Mercantile Trust Co. v. Hensey, 205 U. S. 298, 27 S. Ct. 535, 51 L. Ed. 811, 10 Ann. Cas. 572; U. S. v. Walsh, 115 F. 697 [C. C. A. 2]; General Fireproofing Co. v. Wallace, 175 F. 650 [C. C. A. 8]). In the ease it bar, not only does this not appear, but the sontrary is manifest. It is quite true that, in 3ase of any “dispute or difference” between 1he parties, the decision of the engineer is made final; but no dispute had arisen. Assuming that Storrs gave an unconditional assent, the contract did not therefore make it final; his powers, being so defined, could not be extended by implication. “Expressio unius.” In fact, he never gave an unconditional assent.

It is also true that h s instructions to the contractor were to be authoritative; but he gave none. All he said was that, if what he felt was in fact the ledge, the foreman might go ahead. This was conditional upon something on which he did rot profess to pass. Perhaps he should have done so, but that is immaterial; his failure was not a substitute for his performance. Nor does it make the least difference whether ;he foreman was authorized to speak for the contractor. Storrs’ assent remained conditional none the less, and the contractor was obliged either to persuade him to act on hie own responsibility, or to accept the condition which he imposed. Indeed, as the thing fell out, Storrs had changed the situation noj a bit; in substance, he left it to the contrae ;or to make his own decision. Thereafter the latter’s only course was to use his own judgment or to insist upon a less equivocal commitment; he chose the first.

The last excuse is the final acceptance of the work by Storrs and the board, and the payment by the towns. As we have already said, the contract gave only a limited authority to Storrs. His approval and acceptance were not decisions of disputes referred to him, and could not conclude the towns, if unknown departures from the contract were later discovered. If the point be good at all, it must rest upon general principles of estoppel. But there could be no estoppel after the bridge was completed. It is one thing to approve work as it goes on, and so to induce the contractor to proceed; it is quite another to approve it when it is done. There are grounds enough for an estoppel in the first case, but none in the second. Payments made after completion are irrecoverable for quite another reason; if made with knowledge of the default they are voluntary, and the law sees no reason in justice to recall voluntary payments. In the ease at bar neither *259the hoard nor Storrs knew of the defects when the work was approved and paid for. On the contrary, they supposed both that the piers were on the ledge and were down to the line. Since they were not responsible for the performance in these regards, they remained free to object when the truth was disclosed.

So much for the main issues. We need add little concerning the conduct of the trial. The charge presented the case quite as we understand it, and the requests either added nothing of substance, or should have been refused. The burden was upon the contractor to establish his excuse, once the breach was proved. The defendant got the full benefit of our earlier decision, which was not impaired by taking a special verdict. The evidence admitted was either competent and relevant, or could not have misled the jury in the face of a clear and satisfactory charge.

So far as we can see, the contractor ignored the terms of his contract and substituted what he, perhaps correctly, supposed was everybody’s understanding. It was a natural, but a perilous, course, and, having adopted it, the loss must fall where the words of the contract put it. Contracts are written to avoid such uncertainties, and, however hardly they may bear, we have no choice but to make them the measure of the parties’ obligations.

Judgment affirmed.