BARNES, Circuit Judge:

This is an action for patent infringement of Gilbert Patent No. 2,875,425. Jurisdiction below rests on 28 U.S.C. § 1338(a); here on 28 U.S.C. § 1291.

The patent was held invalid in the district court for lack of invention in that it represents a combination of admittedly old elements 1 with no unexpected and surprising results.

The patent is for a multiple electrical connector or socket designed to receive printed circuit panel boards (containing male insertable devices). The connector consists of a base receptacle and a plurality of spring contacts mounted therein, which make contact with the printed circuits.

The spring contact sought was one that had the following characteristics:

(a) Was of low “contact resistance,” measured in.ohms or voltage drop. The voltage drop is determined by pressure exerted by the contact, and the area of engagement. Both greater pressure, and greater area of engagement (up to a point) decrease voltage drop.

(b) gripped with strength sufficient to maintain the panel boards in place;

(d) was wear resistant;

(e) accommodated various panel board thicknesses;

(f) accommodated panel board distortion, such a warpage, etc.;

(g) resisted shock and vibration;

(h) required minimal insertion force, with subsequent improved grasping force.

These were all important considerations for optimum operation, and were characteristics required by government specifications.

The problem was to combine strength with weakness.. A weaker spring can withstand greater deflection without distortion than can a stiffer spring. A weaker spring will not cause as much wear as will a stiffer spring. A weaker spring has better vibration characteristics. A weaker spring better accommodates differing panel boards. But if too weak, a spring will drop voltage and not grip firmly enough.

Two principal types of springs are here considered, but the parties differ in their respective characterizations of these types. According to appellee, they are “a single-leaf spring contact” as compared with “a more flexible three-leaf spring contact shaped in the form of an S.” (Finding 14, Tr. p. 37.) According to appellant, the two types are an ordinary “leaf spring,” and a “compression spring.” As we read the record (Ex. 27), appellant claims the patented device has the advantage of changing from a leaf spring to a compression spring with the insertion of the panel board. As appellant states:

“As the board moves from point A to point B [on Ex. 28] the spring functions as a leaf-spring, distorting at the bend b, and the force resisting insertion of the panel board slowly increases. When the board reaches *185 point B the action of the spring changes — it no longer functions as a leaf spring, but instead functions as a compression spring — a spring in which the forces exerted on the support and on the inserted board are axially opposed to one another.”

The change from a “leaf-spring” to a “compression spring” allegedly results from the combination of (1) the shape of the spring and (2) the manner of mounting. (O.B. p. 14.)

The shape of the patented spring is “a strip comprising a plurality of sections connected by integral reverse bends to comprise a bellows-shaped spring unit.” 2

The manner of mounting is described as:

“(b) ‘the innermost of said sections being lodged against the inside wall of a pocket’;

“(c) ‘anchored at one end to the receptacle at a point remote from the open end of said trough and extending toward said open trough top’;

“(d) ‘the outermost of said sections defining the contact which is engage-able by a mating contact area of the plug assembly and comprising a length of said strip the free terminal portion of which lies adjacent [to] said open trough top’;

“(e) ‘the surface of said strip length directed toward the interior of said trough being flat’;

“(f) ‘the said bellows-shaped spring unit being contact active as a compression spring’;

“(f) (sic) ‘whereby a full surface engagement is obtained between said contact section of the unit and the mating contact area of the plug assembly.’ ”

(O.B. p. 15.)

The unique feature claimed by appellant for the patented article is the simultaneous achievement of:

(1) Wide surface contact with terminal strips on the board;

(2) sufficient gripping on the board to hold it in place;

(3) an effective electrical connection between the terminal strip on the board and the terminal by projecting down from the connector; and

(4) accommodation of boards of widely varying thicknesses.

Appellee urges (I) that appellant merely took its old or prior commercial connector body (of Ex. 37) and replaced the single leaf spring (which was not sufficiently resilient' and would “set” when overstressed) with an old (in the art) S-shaped spring contact which can flex a greater distance without being overstressed; (II) that even appellant’s expert admitted that once one decided to use an S-shaped spring any engineer or mechanic could design the spring to effect any pressures or degree of movement that he wished by the application of his mechanical skill within the capabilities of the spring. (Tr. pp. 221, 225.) 3 (Ill) that the art related to printed circuit connectors is relatively new, beginning in about 1951; (IV) that the fact S-shaped springs had not been used in printed circuit connectors prior to the birth of the art is of no consequence since others were known to prior art in other types of connectors “since at least 1931”; (V) that at least seven examples of use of S-shaped springs in connectors existed in the prior art (including three foreign patents).

The court, in passing on the matter in dispute, made detailed findings, which appear in the margin. 4

*186 To summarize — the court found the newly patented device produced a questionable superiority in the various areas claimed which was “a mere matter of degree, and does not constitute a new or surprising function or result.” (Finding 31.)

The real question before us in this case is whether the findings are supported by the evidence. (Alleged Errors I and III.) Appellant urges that the findings are not so supported because:

(1) The trade recognizes the marked superiority of the patented construction.

This is arguable, but the evidence is not conclusive, nor does it appear without dispute.

(a) At the time of trial, competing prior art connectors were still being sold. (Tr. pp. 602-11);

(b) Military Specification C-21097 (Ex. 52) issued early in 1958 requiring S-shaped springs was revised (Ex. 57) in 1959 so as not to require S-shaped springs. 5

(c) Sales at higher prices is some evidence of superiority. But we need not agree with appellee’s sug- *187 gestión that the purchasing agent is likely, through an average lethargy, to find that price per se is not controlling.

(2) The patented connector has superseded the prior art connectors.

This again is true, to a degree, but not so much as to render the court’s findings without any substantial evidentiary support. 6

We hold that findings of fact 30 and 42 are sufficiently supported by the evidence. 7

(3) The findings “are merely a one-sided, incomplete, inaccurate document prepared by counsel for defendant with the end in view of supporting only his side of the case.” (O.B. p. 4.)

Little else is ever submitted to a district court judge trying the case, no matter which side wins, and this is particularly true in most patent cases. 8 Of course, this never excuses an exercise by the trial judge of his duty to carefully consider, weigh and determine the aecuracy of the proposed findings, and whether they are supported by the evidence in the record before him.

We assume the trial judge did just that. Appellee states that the court “received and made changes and corrections thereto” (i. e., to the findings). This is factually correct, but the court actually made one notation with respect to the dismissal of the charge as to one patent, one possible correction as to substance (p. 39, 1. 13), and one as to form (separating the Finding and Judgment). All of these actions of the trial judge appear in the proposed findings. (Clk’s Tr. pp. 32-53.) This limited revision of the proposed findings may indicate that ap-pellee’s counsel drew findings so precise and accurate that they needed little change or correction, as well as it might indicate that the findings were blindly followed by the trial judge. It certainly does not prove that the trial judge failed to do his duty to carefully examine the proposed findings, and correct and modify them if necessary.

(4) Appellant charges that the court below expressed an attitude that it “needed no expert testimony” (O.B. p. 55, n. 107, with transcript references cited).

The court below insisted that common sense in evaluating the models of patents was not out of place (Tr. p. 552, 11. 6-9). Yet the court permitted the introduction of expert testimony regarding the attributes of the various models (p. 187, 11. 6 to 22); and listened to all such testimony in the expressed hope it might enlighten him (Tr. pp. 745-46).

No trial judge of any experience has failed at one time or another to figuratively throw up his hands in dismay as he listens to diametrically opposed “expert testimony,” as produced by the two sides. An expert’s testimony is no better than the conviction it produces in the hearer’s mind. Many trial judges are not as frank as the trial judge here, who tells both sides of his opinion of some expert testimony. Such a complaint from the losing side of the lawsuit is as old as the acceptance of expert testimony *188 itself. Sartor v. Arkansas Natural Gas Corp., 321 U.S. 620, 64 S.Ct. 724, 88 L.Ed. 967 (1944); 6 Moore, Federal Practice ¶ 56.15 [4] n. 61; II Wigmore on Evidence § 662 (3d ed.).

(5) Appellant gives no credit to the trial court for the opportunity to observe, handle, experiment, evaluate and draw his own conclusions with respect to the operations of the various connectors, as shown by the various models thereof in evidence. All this is of valid evidentiary value. Appellant’s counsel at the time did not object to such tests. In fact, he approved (Tr. p. 444). Now, he urges it is improper.

While appellant does not dispute the rule that a court, presented with conflicting expert testimony, must choose which is to be believed, it claims error wherein the court “in a case involving technical issues [disregards] the expert testimony and substitutes his own, scientifically uninformed opinion.” (O.B. p. 55.) Such a rule would be valid only where the expert testimony was all uncontradicted and all obviously favorable to the appellant’s position. Such a state of imbalance in the evidence did not here exist. As appellee states in its brief:

“The technical opinions expressed by these two experts differed at times, and apparently the Court below chose to accept the opinion of Dr. Ross, who had actual experience in the field, instead of Professor Siegel, who had only theoretical knowledge of the subject matter involved.” (R. B. p. 21.)

Of course, we recognize that Dr. Ross, appellee’s factory manager since 1960, was not an entirely disinterested witness. But certainly neither was Dr. Martin Siegel, Professor of Mechanical Engineering at the University of Southern California (Tr. p. 80), who had never seen a connector of the type involved in suit until six months prior to the trial when he was retained by appellant as an “expert.” We do not down-grade either expert. We point out only that neither was without interest in the outcome. 9

(6) Appellant also contends that because it made certain tests which allegedly proved what its product could do, and appellee offered no evidence as to tests of what prior art could do, there was a failure of factual proof to support the findings.

We know of no rule which lays down exact limits as to the precise or limited type of factual proof the trier of facts must accept. His own examination of the patent in suit and the prior art may be sufficient. See Graver Tank & Mfg. Co. v. Linde Air Prods. Co., 336 U.S. 271, 274-275, 69 S.Ct. 535, 93 L.Ed. 672 (1949); Acme Steel Co. v. Eastern Venetian Blind Co., 130 F.Supp. 459 (D.Md.) affirmed 227 F.2d 914 (4th Cir. 1955); Philadelphia Co. v. Securities & Exch. Comm’n, 85 U.S.App.D.C. 327, 177 F.2d 720 (D.C.Cir. 1949).

Having disposed of the preliminary matters raised required by the unusual manner in which this appeal was presented to us, we turn to the errors alleged. They are:

I. The findings are clearly erroneous because without support.

II. The findings are clearly erroneous because they fail to refer to a host of clearly relevant facts.

III. The findings are clearly erroneous because they disregard all the credible technical testimony in the case.

IV. Procedural errors or abnormal-ties make the findings suspect, and this court must review them with particular care.

We think we have said enough above to conclude that appellant’s first and third alleged errors are without merit. We hold there was substantial evidenti-ary support for the trial court’s findings, and that he did not disregard all the credible technical testimony. 10

*189 As to appellant’s fourth point, we have reviewed the findings “with particular care.” We are satisfied with them, and as we said above, we find no prejudicial procedural errors.

We do not think the court’s desire (a) to shorten the trial; (b) to hasten the trial so that he could turn to other pressing business; or his action (c) to interrupt the trial for six weeks; (d) to require briefs before all the rebuttal testimony was in — constitutes, either singly or collectively, error in this case. (We cannot fail to speculate whether, under the same circumstances, counsel for appellant would have conceded error had the trial judge ruled in his favor.) Nor do we find (e) that the court substituted his own lay opinion for, and to the disregard of, expert testimony; nor (f) that there is any proof the court “may” have based its opinion upon facts not in the record, “but known to him alone.” 11

Appellant concedes the trial was not “unfair” .in any invidious sense— merely that this court should carefully examine the record. Within the limits of our capabilities (not having the opportunity to observe the witnesses’ demean- or, etc.) and of our responsibilities (to look at the evidence most favorable to appellee), we must ask whether there was sufficient and substantial evidence, which, though disputed, if believed, was sufficient to support the findings. We have conducted this inquiry, and have determined that such sufficient and substantial evidence does exist.

Turning to specification of error two, that the findings make no specific reference to a host of clearly relevant facts, no recital of such facts is attempted in the opening brief, nor is any rebuttal attempted in appellee’s brief. In appellant’s reply brief this issue is first watered down and then evaded. Appellant says (R.B. p. 7):

“The fault we find with the Findings of Fact is not merely that they were adopted by the court below and that they are in large pari unsupported by the proof, but also that they represent a one sided statement [etc.] * * * and in the absence of any opinion by the court we have no way of knowing whether or to what extent they accurately reflect his thinking.”

We can assume, we believe, that any findings signed by a trial judge do reflect his thinking. We can not only assume, but must also presume they accurately reflect his thinking.

But appellant goes on:

“Here we are confident the court will discover that many of the findings have no support in the Record. But apart from that conventional ground (sic) for reversal, we submit that unless this court looks beyond the findings in this case, a situation is presented where a District Court ruling is, for all practical purposes rendered appeal-proof.”

The court is not required to discover or attempt to discover errors that appellant’s counsel cannot or will not point out. Appellee has made specific reference to the evidence in the record supporting Findings 8 to 10, inclusive, 13 to 14, inclusive, and 16 to 43, inclusive. In not one instance has appellant pointed out in its reply brief where such references are inaccurate or untrue. It is true that in its opening brief, appellant attacked all the above findings for inaccuracy or incompleteness, but only ten findings of the 32 (Nos. 24, 32-39, 41) refer to a *190 total lack of proof. We consider each in turn:

(a) Findings 32 to 39, inclusive, refer to the difference between the patent and the prior art as “merely a matter of degree.” Appellee, to support this, cites the tests made in court by the judge; and cites record references to prove both types are used interchangeably in devices for government use, and that the earlier art, like the patent in suit, meets military specifications (Ex. Y). Exhibits 52 and 57 also establish what each connector must do to qualify for governmental specifications.

In rebuttal (R. p. 151), appellant admits that certain supporting testimony exists in the record (an expert’s answer, for example), but that such answer is “meaningless” (R.B. p. 4); and that the appellee’s reference to Exhibit Y (which contained proof that a certain type of construction was being sold) although introduced in evidence constituted “pure hearsay” in stating that the earlier product or art “meets military specifications” (R. p. 5). This is not the stuff from which reversals are earned.

(b) Finding 24 is alleged by appellee to be supported by the transcript, pages 594-96, i. e., “the cooperative relations therebetween are old and well known in the art.” In rebuttal, appellant urges that this was based on the testimony of experts — that “the true and accurate” explanation “was that given by its expert” — that “this court is urged to compare the cross-examination of Professor Siegel with the cross-examination of Mr. Ross and to draw its own conclusion.” (Reply Br., p. 9.) This is not the function of an appellate court. We are not to re-evaluate conflicting expert testimony. Graver Tank & Mfg. Co. v. Linde Air Prods. Co., supra; Carolina Lee Knitting Co. v. Johnson & Johnson, 275 F.2d 91 (4th Cir. 1960).

(c) Number 41 is the final finding alleged to be totally without evidence to support it. It concerns the comparison of contact resistance between the patented contact and prior art contacts. Tests in this area were conducted in open court, answers appellee, citing transcript pages 540-50. But says appellant in his reply brief (a) “our expert later testified these tests were meaningless,” and (b) the finding refers to “prior art” tests, and no “prior art” was tested in open court.

If we concede the reference to “prior art” connectors in Finding 41 is improper, we can only conclude it should be stricken entirely or modified to read “other connectors.” We consider it at most harmless error. Rule 61. United States v. Borden Co., 347 U.S. 514, 74 S.Ct. 703, 98 L.Ed. 903 (1953). 28 U.S.C. § 2111.

In summary we conclude the fourth and last error considered (listed second in appellant’s brief), is without substantial merit.

This court might or might not have agreed with the trial court that on the record before it, the “surprising and unexpected results” test was not fulfilled. There is testimony that the results produced are what would be expected to be produced by one skilled and experienced in this art, young though it may be. Perhaps it is true there had never been the simultaneous grouping of all the results claimed before this particular attacked patent was created. (Siegel’s testimony, Tr. pp. 751-52.) But as ap-pellee points out, the court below put the answer to this question in proper perspective by asking (Tr. p. 752):

“The Court: Now, that answer ‘No,’ what am I to imply from that? That ‘No’ means that none of those elements exist in any of the prior art, or that they do not all exist in one piece of the prior art?

“The Witness: They don’t all exist simultaneously.

“The Court: But in one form, in one way or another some of them exist in the prior art, is that right ?

“The Witness: Yes, sir.”

Cf: Farr Company v. American Air Filter, 318 F.2d 500, 502 (9th Cir. 1963); *191 Griffith Rubber Mills v. Hoffar, 313 F.2d 1 (9th Cir. 1963).

“S”-springs are nothing new. They were known to prior connector art, as were the various placings of the attachments of the spring to the spring holder, or housing. Whether the “S”-spring is a refinement of the leaf-spring, or a different spring, is of little importance. Use of either is not new. Nor do we think we should disturb the conclusion of law made by the court below that the patent in suit was not invention, and that any “invention” was obvious to a person having ordinary skill in the art.

The result we reach is not one based on any belief we hold that we are bound by the conclusion of the trier of fact below as to the validity of a claim in a patent. Validity of a combination patent is determined as a matter of law:

“The mere aggregation of a number of old parts or elements which, in the aggregation, perform or produce no new or different function or operation than that theretofore performed or produced by them, is not patentable invention.”

Lincoln Eng. Co. v. Stewart-Warner Corp., 303 U.S. 545, 549, 58 S.Ct. 662, 664, 82 L.Ed. 1008 (1938).

Here the spring gripped the male insert. Springs had thus functioned before. S-springs functioned to produce a greater area of contact with the insert. They' had so functioned, to a greater or lesser degree, in the prior art. (German, 471,033; Sweden 137,196; German 508,-604; Soreney, 1,880,511.) The housing itself was old in the art. Here the shape or design of the S-spring, and its method and place of attachment, produced a variance in the point at which the greatest area of contact between spring and insert took place. This was also affected by the material used in the spring, by its thickness or thinness, and by the thickness of the insert. This was true, though to a lesser degree, in much of the prior art. (Appellee’s Brief, opposite p. 5 including the Jackson and Kennedy patents.) This was, at best, an improvement in efficiency of function, but not the creation of a new, different or additional function, or a new or different result. It was not unexpected. It was at best the “mincing step forward” described in 69 C.J.S. Patents §§ 204, 213, pp. 686-687, 726-727.

A strict construction of a combination of old elements is required of our courts.

The court below came to its conclusion of invalidity as a matter of law. As a matter of law, we cannot say he was in error in his conclusion that the standard of invention had not been met. See Bergman v. Aluminum Lock Shingle Corp., 251 F.2d 801 (9th Cir. 1957); Kwikset Locks, Inc. v. Hillgren, 210 F.2d 483 (9th Cir. 1954).

We affirm.

. Admitted, that is, by appellant’s expert, Dr. Siegel. (R. 755.)

. To an ordinary Western layman, it might be described as a “lazy-S” shaped spring.

. The second of these references to the transcript only partly supports the conclusion stated by appellee, i. e., “any pressure.”

. Finding SI: “The plaintiff contended at the trial that the patented connector is superior because the use of the S-shaped spring contact and the flat-surface con *186 tact of the spring contact with the panel hoard resulted in the following advantages: (a) the contacts are more flexible and will permit panel boards of varying thicknesses to be received by the connector, (b) the flat-surface contact causes less wear on the panel board, (c) the large-surface contact causes a firm retaining grip on the board, (d) the large-surface contact will hold well under shock and vibration, the large area of contact will provide least electrical resistance, (e) the greater flexibility will permit a large number of insertions of a panel board into the connector, and (f) the greater flexibility will increase the ease of insertion of a panel board into the connector. However, there is no evidence that the prior art connectors and particularly the Kennedy-type connector were unsatisfactory in regard to any of the above particulars and, indeed, the evidence is to the contrary in that the Kennedy-type and patented connectors are used interchangeably in computers built for the government (Exhibit BJ). Even if it were assumed that the patented connector was superior in the above particulars, such superiority is a mere matter of degree and does not constitute a new or surprising function or result.

Finding 32: “Both the patented connector and the prior art Kennedy connector have flexible spring contacts. Any difference in flexibility of the spring contacts is merely a matter of degree.

Finding S3: “Both the patented connector and the prior art Kennedy connector are designed to accept printed circuit panel boards of varying thicknesses. Any difference in amount of variation in panel board thickness accepted by each connector is merely a matter of degree.

Finding 37/: “Both the patented connector and the prior art Kennedy connector will cause wear on a panel board inserted thereinto. Any difference in amount of wear on a panel board produced by these connectors is merely a matter of degree.

Finding 35: “Both the patented connector and the prior art Kennedy connector retain the inserted panel board by the grip between the spring contacts and the panel board. Any difference in gripping of the board by these connectors is merely a matter of degree.

Finding 36: “Both the patented connector and the prior art Kennedy connector will hold panel boards under shock and vibration without electrical failure. Any difference in ability of these • connectors to withstand shock and vibration is merely a matter of degree.

Finding 37: “The contact resistance of the patented connector and the prior art Kennedy connector is substantially the same, and any difference of contact resistance therebetween is merely a matter of degree.

Finding 88: “Both the patented connector and the prior art Kennedy connector will permit a large number of insertions of panel boards into the connectors without failure. Any difference in the number of insertions of panel boards into these connectors is a mere matter of degree.

Finding 89: “The ease of insertion of a panel board into a patented connector is substantially the same as the ease of insertion of a panel board into a prior art Kennedy connector, and any difference in ease of insertion is merely a matter of degree.”

. Appellant urges this revision only resulted because the government does not wish to specify patented devices. Appel-lee urges it is much more likely that the government recognized non-S shaped *187 springs could produce the same results. Neither position urged is established by adequate proof. We need not decide the reason for the government’s change in position; we are interested only in the fact of its change.

. Amyphenol-Borg is the largest connector manufacturer in the United States. (Ex. 50.) It does not use the S-spring in its connectors.

. “30. The prior art tuning fork and Kennedy connectors are still being manufactured, sold and used, and have not been superseded by the patented device. Hi H: Hi »

“42. There are many types of printed circuit panel board connectors on the market presently, of which the patented connector is merely one form, all of which are designed to do the same job and which do perform the same job in substantially the same way and obtain substantially the same results.”

. The findings proposed by appellant in this case (Tr. pp. 80(M)5) are not before us.

. Professor Siegel, preparing to testify, made no study of prior art. (Tr. pp. 181-82.)

. We need not reach the question whether the court may disregard all technical testimony in a patent case.

. Appellant makes reference to “vaguely described springs which the Court remembered, but with respect to which there was no proof in the record.” Appellant cites accurately pages 97, 201, 215 and 216. All this was but a pleasantry made by the court in passing with respect to springs holding windows, or desk drawers, in the position in which they had been placed, as seen by him in his youth. There is no merit to the contention he relied on this knowledge or recollection to the exclusion of the evidence introduced in this case.