This damage suit for breach of contract was tried to the court without a jury. It resulted in a judgment for defendant-appellee, hereafter Ross Island.

The appellant (hereafter Beck), raises the following questions:

(1) Did the trial court err in Finding of Fact XI and Conclusion of Law I that the Standard Specifications for Ready Mix Concrete, C94, was made applicable to the concrete here in question by the specifications in the construction contract between Beck and the owner of the building?

(2) Was ASTM C94, 1967 edition, improperly relied upon by the trial court?

(3) As to the sufficiency of the evidence and the court’s findings, Beck contends the concrete, supplied by Ross Island, failed to meet compressive strength standards. As to these findings, three claims are made by Beck as to the lack of compressive strength:

I

Finding XI and Conclusion I

Beck attacks Finding of Fact XI stating that the standards of ASTM “Standard Specifications for Ready Mix Concrete, C94” were made applicable to the concrete in question by the “construction contract specifications.” Attack is also made on Conclusion of Law I referring to the same contract as the “construction contract between plaintiffs and the building owner.” Beck points out that none of the contract betweén Beck and the building owner was incorporated into the contract between Beck and Ross Island; that the contract between Beck and Ross Island consisted solely of the purchase order dated December 30, 1965, which was Exhibit “B” to the pre-trial order, and only required that Class B-l concrete achieve a compressive strength of 3,000 psi at 28 days after pour.

We think that Beck is correct in that the court was mistaken in its reference to the construction contract between Beck and the owner of the building and that there was nothing in the purchase order, Exhibit “B” to the pretrial order, which made reference to the standards of ASTM. However, we think there was no error in this respect and that the *319 contention is an after-thought on the part of Beck.

Our reasons are as follows:

The contention was never raised at any time until the filing of Beck’s opening brief. In the motion for new trial, etc., Beck’s counsel, Mr. Nahstoll, stated that the 1964 edition of ASTM C94 “was the edition applicable to the terms of this contract;” and Mr. Nahstoll stated in his affidavit supporting his motion for a new trial as follows: “The 1964 edition of ASTM-C94 which was in effect and applicable to the contract between plaintiff and defendant during the time the defective concrete complained of in this case was supplied by the defendant. . . .” Beck also offered in evidence certain portions of ASTM C94.

Thus, even in the motion for a new trial the contention was not raised and was raised for the first time in Beck’s opening brief. The parties having tried the case as if ASTM C94 was part of the specifications to be followed, Beck may not now raise this contention on appeal.

II

The use of ASTM C9k, 1967 Edition

The concrete was poured in July 1966. The trial court relied on ASTM C94, 1967 edition. However, the fact that it was a 1967 edition was not brought to the trial court’s attention until raised by appellant on motion for a new trial.

The 1967 edition was used in the trial, and although apparently no date appeared on it, appellant’s counsel could have ascertained its date. In fact, appellant placed part of ASTM C94, 1967 edition, in evidence.

The motion for a new trial, etc. set forth the 1964 and 1967 editions of ASTM C94 and was supported by counsel’s affidavit that after trial he had ascertained that a 1967 edition had been used and relied upon.

The question of whether the 1964 edition was newly discovered evidence within the well-known rule, and a new trial should have been granted, was one for the court’s discretion. We find no abuse thereof.

III

The concrete delivered to the fifth floor

The claim as to the fifth .floor was for “delay damages.” The concrete was not replaced. Beck ceased work after a 7-day test. It contends that the concrete did not harden until 42 days, instead of 28 days called for in the contract.

We hold that there was sufficient evidence to support the trial court’s findings that the Class B-l concrete poured on the fifth floor satisfied the compressive strength requirements of the purchase order.

Moreover, correspondence beginning August 31, 1966 and invoices from appellant to October 26, 1967, which purported to contain an itemization of Beck’s damages, made no reference to the fifth floor. The complaint filed December 5, 1967, makes no claim for the fifth floor. The claim was finally asserted in late 1968, two years and 17 depositions after the event.

IV

The class A-l Concrete for girders on the fourth floor

The Class A-l Concrete was delivered July 8, 1966, for the girders on the fourth floor. When the slab was determined to be defective, Beck insisted that the girders as well as the slab be removed. The girders were removed over Ross Island’s objection. The girders were sound.

There was conflicting testimony as to whether the girders needed to be removed and repoured, because of a lack of bond to the steel bars on the fourth floor slab.

The trial court did not err in finding the evidence failed to establish a defect in the girders underlying the fourth *320 floor slab, and thus impliedly finding that the bond was not defective.

V

The Class B-l concrete in the fourth floor slab

The main thrust of appellant’s argument at the hearing in this court and in its briefs involved the concrete poured in the fourth floor slab. Appellant Beck attacks the sufficiency of the trial court’s findings as to the fourth floor slab. Ross Island supports the findings and contends that (1) Beck did not prove the concrete, as delivered, would not have met the 3,000 psi requirement in 28 days; (2) that if the concrete was defective, it was caused by Beck’s addition of excess water; and (3) that Beck assumed the risk of the use of Pozzolith in the mix.

The purchase order provided:

“Class B-1; 3,000 psi at 28 days ¾" aggregate, 6 sack cement with 3L poss.” 1

Agreed Fact No. 20 in the pretrial order reads:

Exhibit “F” lists various concrete mixtures, and states:

“These quantities will produce about a 4" ‘slump.’ ” 2

We conclude that the purchase order and Agreed Fact No. 20 provided that Ross Island agreed to supply concrete mix (1) with a compressive strength of 3,000 psi at 28 days after pour and (2)' with approximately a 4" “slump.”

As to the concrete in the fourth floor slab, the trial court made one general or ultimate finding, No. XXII, and several more specific findings, Nos. XVII, XIX, XX and XIII, which had a bearing on liability. We discuss each hereafter.

The sufficiency of an ultimate finding of fact, which alone may purport to decide a case, has been questioned. Federal Practice and Procedure, Rules Edition, Barron and Holtzoff, Wright; § 1127 (1961). See Kelley v. Everglades Drainage District, 319 U.S. 415, 422 , 63 S.Ct. 1141 , 87 L.Ed. 1485 (1943) and Official Creditor’s Committee of Fox Markets, Inc. v. Ely (9 Cir. 1964) 337 F. 2d 461, 467 .

We do not intend to retry the case or to undermine the well-accepted rule in this circuit that findings of fact by trial courts in civil cases will not be set aside unless clearly erroneous.

We recognize that “if the findings are sufficient to support the ultimate conclusion of the court they are sufficient .... The ultimate test as to the adequacy of findings will always be whether they are sufficiently comprehensive and pertinent to the issues to provide a basis for decision and whether they are supported by the evidence.” Rayonier Inc. v. Poison (9 Cir. 1968) 400 F.2d 909, 923 .

We may set aside findings, even if there is some evidence to support them, if we have a firm conviction that a mistake has been made. Where a case is tried to a court without a jury, “A finding is ‘clearly erroneous’ when although there is evidence to support it, the reviewing court on the entire evidence is left with the definite and firm conviction that a mistake has been committed.” United States v. United States *321 Gypsum Co., 333 U.S. 364, 395 , 68 S.Ct. 525, 542 , 92 L.Ed. 746 (1948). Accord: Smith v. James Irvine Foundation (9 Cir. 1968) 402 F.2d 772 .

We think that the nature of the findings as to the fourth floor slab requires a reversal as to this part of the decision. Accordingly, we reverse that portion of the judgment and remand for a new trial.

Facts Proven

Fifty cubic yards of Class B-l concrete was delivered by Ross Island for the fourth floor slab on July 8, 1966. Each truck load carried 7 to 8 cubic yards of concrete. The evidence is un-contradicted that the Pittsburgh Testing Laboratory cement inspector found the loads “unusually stiff” in the trucks and that on two of the trucks the concrete would not pour out of the truck. Water was added by Beck’s employees to get the concrete in proper “slump” for pouring. Ross Island’s expert, Polivka, testified that this unusual stiffness of the concrete indicated misbatching or some other defect in the condition of the concrete. The last load had an unusual brownish appearance and crumbled out of the truck.

Within three days after the southern portion of the fourth floor was poured, it was apparent that the concrete was defective. On that day, July 11, the concrete inspector of Pittsburgh Testing Laboratory noticed the concrete in one of the testing cylinders 3 had an unusually white appearance and crumbled easily. He notified two of the job supervisors from Beck and they went up to check out the fourth floor slab. Using a nail to scratch the surface, they found all the slab poured on July 8 to be soft. It had not set up properly.

Mr. Galloway, vice president of Ross Island, was immediately notified. He came over to the job site and personally inspected the slab. He then suggested that two of the test cylinders taken during the July 8 pour, be steam cured by Pittsburgh Testing Laboratory and their compressive strengths be measured the next day. Galloway thought that the steam curing would give an approximation of the compressive strength of the slab 28 days after pour. The cylinders were steamed cured and strength tests were made on July 12. They registered strengths of 650 and 350 psi. Upon learning the results of the test, Galloway agreed that the fourth floor pour of July 8 should be torn out and repoured, and that Ross Island would be responsible for the cost. This work commenced immediately on July 12. The oral agreement of July 12 was confirmed by a letter from Galloway to Beck, dated July 14. 4

After the concrete was removed, a sample of it was sent to Master Builders *322 Company, Cleveland, Ohio. They analyzed the concrete and concluded that it contained three times more Pozzolith 3L than the contract specified. They reported that the concrete contained 6.5% more air than if the correct amount of Pozzolith 3L had been used.

Master Builders also reported the cement content was well below the contract specifications. At the trial, Ross Island’s expert, Polivka, conceded there were only 4.6 sacks of cement, per cubic yard, instead of the 5 sacks required by the contract. This confirmed the admission in Galloway’s letter that the batching equipment used by Ross Island in weighing cement was defective. 5 Ross Island had also received complaints about the quality of cement from two other customers on July 6. Thus the evidence showed that the concrete arrived at the job site on July 8 in a mis-batched condition.

• In addition to the two cylinders, steam cured and broken on July 12, which showed readings of 650 psi and 350 psi respectively, referred to above, another cylinder was broken and tested by Pittsburgh Testing Laboratory on July 15, 7 days after pour. The concrete from this cylinder tested 710 psi, whereas concrete tested at this point should have shown 45 to 60 percent (1,350-1,800 psi) of its 28-day strength of 3,000 psi. Finally, a test cylinder from the July 8 pour was broken by Pittsburgh at 28 days after pour and showed 1,700 psi instead of the 3,000 psi required by the purchase order, or based on ASTM C94’s 79% figure, 2,370 psi.

The trial court made the following finding of fact, No. XII, its major finding as to the fourth floor slab:

As shown by the facts set forth above, we have a firm and definite conviction that a mistake was made and hold the finding was clearly erroneous. United States v. United States Gypsum Co., supra.

The trial court also made finding of fact No. XX, that the test of the cylinder broken at 28 days after the pour showing 1,700 psi, “is not sufficient proof that even the concrete in that truck load [from which the cylinder was taken] was of insufficient compressive strength.”

We hold this finding was also erroneous. It should be noted that the trial court had sustained the quality of the fifth floor cement on the basis of a single 28-day test which satisfied the compressive strength requirements. Moreover, the single 28-day test on the fourth floor concrete came from the first truck load delivered on July 8 (Finding No. XX). It was the last four of seven truck loads which showed the most obvious defects in the batching. It is a reasonable inference that the last loads would have shown even less compressive strength.

Finding of Fact XVII reads: “The addition of water to the concrete delivered on July 8, 1966, in the amounts previously indicated, was inconsistent with accepted practice in the industry.” However, we hold the finding was clearly erroneous. The record clearly supports the conclusion that adding water to concrete at the job site was a customary practice. A representative of Pittsburgh Testing Laboratory, a vice president of Ross Island, and a professor from the University of California at Berkeley, who was an expert witness for Ross Island, all testified that adding water at the job site was a customary practice. Also, on the trip ticket forms that *323 the drivers for Ross Island used, there was a space provided for the driver to indicate how much water was added at the job site.

As to the possibility of the addition of too much water, the trial court made Finding of Fact No. XIX:

This finding does not aid Ross Island. It did not find excessive water or excessive Pozzolith was added. The record is undisputed that three times the amount of Pozzolith called for in the purchase order was found in the poured fourth floor cement.

The record shows without dispute that the “slump” in the loads delivered for the fourth floor did not nearly approach the approximate 4-inch “slump” called for in the contract; 6 and that Beck followed customary practice and acted reasonably to mitigate damages in adding additional water to make the concrete workable. 7

Finding of Fact No. XIX correctly states that Ross Island “did not contract to meet any seven day requirement.” But the 7-day test was only one of various tests made on the fourth floor concrete and like the other tests, was one means of ascertaining whether the concrete would meet the 28-day test. We conclude that Finding of Fact No. XIX does not prevent our reversal of that portion of the judgment involving the fourth floor slab.

Conclusion of Law No. IV reads:

The conclusion that Beck assumed the risk of the use of Pozzolith is erroneous. If Beck assumed a risk, it was a risk of use of concrete with a proper amount of Pozzolith and not, as the record shows without dispute, three times the contract’s amount of Pozzolith.

Conclusion

We cannot escape the fact that Galloway, an officer of Ross Island, observed the concrete after it was poured on the fourth floor, conferred with Beck’s officials and orally agreed that the concrete on the fourth floor should be removed and replaced at Ross Island’s expense. In addition, he confirmed this by letter. The trial court made a finding, No. XXIII, to the effect that Galloway’s letter was written before he knew or should have known of the results of the cement concrete tests or of the large addition of water. The finding concludes, “Notwithstanding this, the plaintiffs did not rely upon this letter.”

The first part of the finding is probably correct as far as it goes, but Galloway knew at the time of his conference and at the time of writing the letter *324 that the concrete had been misbatched. The last sentence of the finding quoted above concerning reliance is hard to explain. We cannot find that it was proposed in any of the proposed findings submitted by Ross Island, nor can we ascertain from the record any evidence upon which the finding was based. Beck was not urging estoppel. In any event no estoppel was found.

Moreover, an inspection of Galloway’s testimony does not assist us in the matter. Whether Beck relied or not, the oral conversation and the letter were admissions of liability.

It is difficult to understand how with this admission of liability as to the fourth floor, Beck could fail in its action as to the fourth floor. We can only speculate that had Beck centered its case on the fourth floor concrete and not tried to drag into the case every possible element of damage in other respects, that a different result might have occurred.

We reverse the judgment as to the fourth floor slab and remand for a new trial on that portion of the case.

VI

Ross Island’s contention that Beck sustained no damage since it had been reimbursed by the building ovmer.

We do not reach this question. The trial court stated that that supplemental fact finding on the relations between Beck and the building owner, would be necessary if Beck recovered damages. Because of the outcome of the case the trial court never reached the issue.

On remand, if Beck recovers damages on the fourth floor issue, the trial court will try the reimbursement issue and make findings of fact and conclusions of law.

Third Party proceedings were segregated for separate trial by order of the trial court. If recovery is had against Ross Island, our decision does not affect Ross Island’s claims as Third Party Plaintiff against various Third Party Defendants.

Affirmed in part; reversed and remanded in part.

. Pozzolith 3L is a lignin and by-product of the lumber industry. It is added to concrete to give it better workability with a lesser quantity of water. An excess of Pozzolith 3L will cause the air content of concrete to increase. This in turn will reduce the compressive strength of the concrete.

. “Slump” is an industry term for the consistency or workability of concrete. A one-inch “slump” would be thick and hard to pour. A seven-inch “slump” would have a soupy consistency.

. A sample of concrete in cylindrical form is taken from various loads throughout the day. These cylinders are considered to be representative of the concrete poured that day. When the strength of the concrete needs to be determined during the hardening period, a cylinder can be subjected to a pressure test to determine at what point it will break. This results in a strength measurement in pounds per square inch.

. “Beck-Utah

P. O. Box 12125

Holladay Park Station

Portland, Oregon

Attention: Mr. John Diekmann

Gentlemen:

“Our firm will accept the responsibility for the cost of removal and replacement of the slab concrete (poured July 8th, 1906) which was found defective on the fourth floor slab of the Calaroga project in the southwest section of the building between lines two through five, and A through a mid-point between lines O and D.

“A mechanical malfunction was found by a factory service man on our Selee-tron batching equipment which caused occasional error in the weighing of the cement. This has now been corrected and we have every reasonable expectation of good performance in the future. Very truly yours,

Boss Island Sand & Gravel Co.

/s/ W. I-I. GALLOWAY Vice-President, Sales”

. There was testimony from Ross Island’s expert that calculations as to the air content, brought the cement up to 4.98 sacks per cubic yard. However, Master Builders’ report confirms that the batching equipment was not batching the amount of cement that Ross Island intended to place in the mix.

. The requirement of a specified “slump” of approximately four inches (Agreed Fact No. 20) insured the workability of the concrete for the purpose of pouring and finishing. Since it was provided for in the contract, a four-inch “slump” would not destroy the projected compressive strength.

. We have discussed, supra, the customary practice to add water to secure a proper “slump.” Beck had a duty to mitigate damages and should have used the concrete delivered, if it could make it workable by adding water and not increasing the “slump” beyond approximately four inches.

Ross Island must be charged with reasonably foreseeing that Beck would so act.