L. HAND, Circuit Judge.

Feldman, through whom the plaintiff takes title, carried four policies of robbery insurance upon his stock of jewelry in New York City, in the sum of $2,000, $5,000, $10,000 and $58,000, respectively. He asserted that two robbers entered his shop in the morning, overcame two employees then present, and made away with some of his stock. Various issues were litigated at the trial, and the jury eventually after considerably deliberation brought in a general verdict of $10,700. The plaintiff thereupon moved for a new trial because the verdict was for inadequate damages, which the judge denied. The only question which can here be raised is whether this order was wrong, and this is based upon the theory that the evidence was such as to require either larger damages or none at all.

The loss depended upon calculations made from Feldman’s books, and the plaintiff made his ease by showing what jewels Feldman had bought, what he had sold, and what remained after the robbers left. The missing stones and settings, so computed, the plaintiff took at their cost, supplemented by Feldman’s own oath that jewelry had not fallen in value since he bought those stolen. There were discrepancies in the books, proving as to some, though not many, items, that he had omitted to record sales, and they were not in general *464 kept very regularly. Nevertheless, the amount of the loss as proved by this evidence was over $51,000, for which .the plaintiff asked recovery. After going out the jury came back for further instructions, and during the ensuing colloquy one of them asked whether they might bring in a compromise verdict. The judge told them that they must find the loss only from the evidence, and eventually they agreed upon the figure stated, though it seems reasonably clear that this was the result of a compromise.

We do not find it necessary to state the evidence more in detail, because we have no power to review the order under these circumstances. Having reached their conclusion that the defendant was liable, the jury had not ended its work. The action was not upon a note, or for a penalty, nor was there even a minimum set by law, or conceded, greater than the verdict. The plaintiff had to satisfy them as to how many jewels Feldman had lost and what was their value; the defendant admitted none of these things, and 'the jury was not obliged to take all the plaintiff’s proof, or reject it all. They might not believe that the books correctly stated the cost, or that the cost was the value. They might think that the irregularity of the boobs made them an unsafe guide to determine what jewels had been stolen. Perhaps, if they had thought these things, they should have found for the defendant, on the ground of fraud, or for breach of warranty, but we are not concerned with the logical consistency of their verdict, as the plaintiff throughout assumes. The question before us is not the same as that before the trial judge, who must treat all these considerations as pertinent. At least when there is no limit fixed by law, or conceded, and when the jury has fixed more than nominal damages, we cannot review an order denying a new trial, the only act of the judge which can be challenged.

At common law a writ of error searched only the record, that is, the judgment roll, made up of the process, the pleadings, the minutes of the clerk, the verdict, and the judgment. It was only by the Statute of Westminster II, (1285), that anything which occurred at the trial could become part of it (Holdsworth, vol. I, pp. 222-226). That statute gave a party the right to a bill of exceptions, as we now call it, which came up with the writ and could be assigned as error. A motion for a new trial, though very ancient, was never regarded as part of the bill, but rather as an independent method of review (Stephen on Pleading, § 99), and there was no review of that review. The Supreme Court in a great many eases, of which a number are stated in the margin, has so decided, at times basing its decision upon this procedural ground. 1 At times it has also added that the matter was in the discretion of the trial judge, and apparently for this reason, in Mattox v. U. S., 146 U. S. 140, 13 S. Ct. 50, 36 L. Ed. 917, an exception came to be grafted upon the rule, when it appeared that the trial court had refused to hear the motion at all. The procedural anomaly seems to have been ignored, and at any rate the exception became established, though it has not been carried further. In Southern Ry. Co. v. Bennett, 233 U. S. 80, 34 S. Ct. 566, 58 L. Ed. 860, for example, the case was as strong to reduce the verdict as here it is to raise it, for the court assumed that by no possibility could the damages have been so high. Holmes, J., indeed there intimated that, if the jury had exceeded a statutory limit, the question might be open, but in general the court has often declared itself against reviewing the damages (Railroad Co. v. Fraloff, 100 U. S. 24, 31, 25 L. Ed. 531; Wilson v. Everett, 139 U. S. 616, 11 S. Ct. 664, 35 L. Ed. 286; Lincoln v. Power, 151 U. S. 436, 438, 14 S. Ct. 387, 38 S. Ct. 224), and, of these, Wilson v. Everett was an extreme case, much stronger than that at bar. So far as. we can find there is no exception to the doctrine in the highest court.

The practice of this court has been uniform to refuse to consider any orders denying new trials, with the exception of Harrison v. U, S. (C. C. A.) 7 F.(2d) 259, which followed Mattox v. U. S., supra. New York & T. S. S. Co. v. Anderson (C. C. A.) 50 F. *465 462; Denison v. Shawmut Mining Co. (C. C. A.) 159 F. 102; Reader v. Haggin (C. C. A.) 160 F. 909; Worden v. Kenny (C. C. A.) 239 F. 131; Ford Motor Co. v. Hotel Woodward Co. (C. C. A.) 271 F. 625; Harrison v. U. S. (C. C. A.) 7 F.(2d) 259. The hooks are full of similar cases in other circuits too numerous to cite, in many of which the reason given was that the order was discretionary, and which apparently implied that there might be a review if the aggrieved party could show that the discretion had been abused. Acting upon this, in several cases orders refusing new trials because of the amount of the verdict have been in fact reversed, though apparently in ignorance of the limitations in the dictum in Southern Ry. Co. v. Bennett. And so the gloss has become the supposed principle, and there has undoubtedly grown up a notion that in general there may be extreme cases where this is a good ground for appeal. However, so far as we have found, all the actual decisions except one, when this has been done, are either'where the jury gave nominal damages, that is, where they did not attempt to appraise the plaintiff’s loss at all; where their verdict was less than the amount of the loss which the defendant did not dispute; or where damages were not involved. In the first class is Pugh v. Bluff City Excursion Co., 177 F. 399 (C. C. A. 6). In the second are Glenwood Irrigation Co. v. Vallery, 248 F. 483 (C. C. A. 8); United Press Associations v. National Newspapers Association, 254 F. 284 (C. C. A. 8); and Stetson v. Stindt, 279 F. 209 (C. C. A. 3). In the third are James v. Evans, 149 F. 136 (C. C. A. 3), and Frye v. Lyon, 299 F. 926 (App. D. C.). All these are within the limits suggested in Southern Ry. Co. v. Bennett, and the exception, Cobb v. Lepisto (C. C. A.) 6 F.(2d) 128 (C. C. A. 9), so far as we can find, stands alone.

That it might be desirable to give such an appeal we do not deny. The direction of, or refusal to direct, a verdict is reviewable, and the question which arises is precisely the same as that upon a motion for a new trial, as-the Supreme Court has declared again and again. Gunning v. Cooley, 281 U. S. 90, 50 S. Ct. 231, 74 L. Ed. -. The Seventh Amendment is no bar to this, nor is it to the trial judge’s “re-examination” of the damages. It is only a procedural difficulty that prevents our review; that is, that the court’s action must of necessity await the verdict. That may not be a good reason, when the matter is of substance, but Congress has been consciously conservative. The Act of April 26, 1928 (28 USCA § 861b) expressly confined the newly given appeal by all the statutes previously regulating the old writ of error, among others by those “defining the relief which may be had.” The predecessors of section 879 of title 28 USCA had from the outset limited the office of the writ by the common law, and the decisions of the Supreme Court had. repeatedly so construed it, as we have shown. These were written into its text, and the recent reaffirmation must be understood as carrying over the settled practice as it was. Hecht v. Malley, 265 U. S. 144, 153, 44 S. Ct. 462, 68 L. Ed. 949. So far as the decisions of the Supreme Court have gone, we can go; perhaps we should go as far as the cases we have cited in the Circuit Courts of Appeal, though that we have not now to decide. But we do not see how wé can follow Cobb v. Lepisto without ignoring settled rules which courts must observe, whatever their disposition were the matter tabula rasa.

Moreover, while it might be well to give us the unconditional power to review such orders, we could not as things are exercise it freely. The most that has been said is that we may do so when there has been “an abuse of discretion.” That is an impracticable rule, unless confined as it has always been confined save for Cobb v. Lepisto. The trial judge decides what verdict is within the bounds of reasonable inference from the evidence. That is a question which we can consider as well as he, and which we do upon his direction of a verdict. But by hypothesis we are not to be allowed to do that; we must come at the matter at one remove, and apply the same test to the judge’s decision that he applies to the jury’s. We must in effect decide whether it was within the bounds of tolerable conclusion to say that the jury’s verdict was within the bounds of tolerable conclusion. To deeide cases by such tenuous unrealities seems to us thoroughly undesirable; parties ought not to be bound by gossamer strands; judges ought not to engage in scholastic refinements. Until the practice is frankly changed, we must adhere to it as it'exists.

The state practice and decisions upon appeals are not relevant. Camp v. Gress, 250 U. S. 308, 317, 318, 39 S. Ct. 478, 63 L. Ed. 997.

Judgment affirmed.

Notes

Henderson v. Moore, 5 Cranch, 11, 3 L. Ed. 22 (1809); Marine Ins. Co., v. Young, 5 Cranch, 187, 3 L. Ed. 74 (1809); Barr v. Gratz’s Heirs, 4 Wheat. 213, 220, 4 L. Ed. 553 (1819); McLanahan v. Universal Ins. Co., 1 Pet. 170; 183, 7 L. Ed. 98 (1828); Parsons v. Bedford, 3 Pet. 433, 448, 7 L. Ed. 732 (1830); Zacharie v. Franklin, 12 Pet. 151,153,9 L. Ed. 1035 (1838); Brown v. Clarke, 4 How. 4, 15, 11 L. Ed. 850 (1846); U. S. v. Hodge, 6 How. 279, 281, 12 L. Ed. 437 (1848); Doswell v. De La Lanzo; 20 How. 29, 32; 15 L. Ed. 824 (1857); Warner v. Norton, 20 How. 448, 461,15 L. Ed. 950 (1857); Pomeroy’s Lessee v. Bank of Indiana, 1 Wall. 592, 597, 598, 17 L. Ed. 638 (1863); Freeborn v. Smith, 2 Wall. 160, 176; 17 L. Ed. 922 (1864); Kerr v. Clampitt, 95 U. S. 188, 24 L. Ed. 493 (1877); Newcomb v. Wood, 97 U. S. 581, 24 L. Ed. 1085 (1878); Holder v. U. S., 150 U. S. 91, 14 S. Ct. 10; 37 L. Ed. 1010 (1893); Moore v. U. S., 150 U. S. 57, 14 S'. Ct. 26, 37 L. Ed. 996 (1893); Blitz v. U. S., 153 U. S. 308, 14 S. Ct. 924, 38 L. Ed. 725 (1894); Addington v. U. S., 165 U. S. 184,17 S. Ct. 288, 41 L. Ed. 679 (1897); Clune v. U. S., 159 U. S. 590,16 S. Ct. 125, 40 L. Ed. 269 (1895).