(after stating the facts as above). This case presents the question whether, in the case of a cargo, insured under a valued policy, which has suffered a general average contribution, the in*679sored is coinsurer to the extent that the sound value of the cargo is more than the agreed value. The record does not show whether the cargo was undervalued at the port of lading, or whether on the voyage it increased in value; for this reason we ignore the first possibility, without suggesting that it makes any difference. The issue, as the books repeatedly say, is of no grave consequence, once the rule is settled, for the parties may provide for it by a clause in the policy, but it is “tabula rasa,” there being no authority controlling upon us, and the practice of underwriters being largely in dispute, at least since the decision of the Ninth Circuit in British, etc., Co. v. Maldonado & Co. (C. C. A.) 182 F. 744.

It has been settled law since 1761 that, in the case of the partial loss of a valued cargo, the insured is a coinsurer in the event of increase in value, and conversely that, in the case of a decrease, he recovers that proportion of the agreed value which his loss bears to the sound value. Lewis v. Rucker, 2 Burr. 1167. This has been generally followed since that time. Johnson v. Shedden, 2 East, 581; Turner v. Edwards, 12 East, 488; Lawrence v. Insurance Co., 3 Johns. Cas. 217; Forbes v. Mfrs.’ Ins. Co., 1 Gray (Mass.) 371. It was recognized obiter in London Assurance Co. v. Companhia, etc., Co., 167 U. S. 149, 172, 17 S. Ct. 785, 42 L. Ed. 113, and in International Nav. Co. v. Atlantic Mut. Ins. Co. (D. C.) 100 F. 317, 318, affirmed on Judge Brown’s opinion in 108 F. 987 (C. C. A. 2). The doctrine seems to us to be in accord with sound principle.

The probability of loss does not vary with the value, the goods being the same; but, the greater the value, the greater the risk. In the light of this, it is certainly possible to read the insertion into a policy of an agreed value as intended to protect the underwriter against increases or undervaluation in eases of partial losses, where alone the question is important. The issue is whether it must be so understood, or whether it may merely limit the amount of the total recovery. Confessedly to the amount of the agreed value not covered the insured is a co-insurer, just as he is in the case of an open policy to 'the extent of the uncovered prime cost at the port of lading. But the agreed value merely stands in the stead of prime cost, as Lord Mansfield said in Lewis v. Rucker. In an open policy, however, the loss is always calculated upon the proportion of prime cost to sound value. Usher v. Noble, 12 East, 839; Clark v. United, etc., Co., 7 Mass. 365, 5 Am. Dec. 50. Unless the agreed value has the same effect as prime cost pro tanto, the underwriter would therefore be at a disadvantage in a valued policy. Hence an agreed value is more than a mere limit upon the recovery; it assures the' underwriter against increases in value, just as it assures the insured against decreases.

Such difficulties as have arisen came from the notion that the agreed value is absolute for all purposes, a “mutual estoppel” against both sides. This is true in general, but it is impossible to understand it as meaning that the parties may upon no occasion use any other value. Except in the event of the destruction of part of a mass of fungibles, the insured cannot establish his loss at all, save by recourse to sound value. Nobody has therefore ever suggested that he is limited to the agreed value, and the result would be preposterous if he were, because he might be able to show no loss whatever, though he had lost much. Hence he fixes his loss at the difference between sound value and the proceeds from the sale of the damaged goods. There can therefore be no “mutual estoppel”; the insured is not preeluded from going outside the agreed value, and the underwriter must also be free to do the same. Only so can he get the benefit of the agreed value, except as a mere limit on the recovery.

So far as we can see, every consideration which dictates this way of reckoning partial losses likewise applies tt) general average contributions by cargoes. The libelant, however, argues that such a contribution is different, because it is fixed in dollars, and the insured needs no recourse to sound value to establish it. That is true, but only because the sound value has already gone into the computation when the adjustment was made, which is conclusive upon the parties. But that cannot conceal the fact’that it has been used, nor does it throw any light upon whether the underwriter’s liability should be different. In faet, the losses are of the same nature, and should be subject to the same rules. The law of the sea makes the interests of all parties to the common venture security for each, so that, when one is injured, the goods of the rest become bound pro tanto to make good the loss. The lien so arising may be foreclosed, the sound goods seized and their proceeds taken, and the loss is spread over all, exactly as though each had been directly injured pro tanto. Were the guarantor entitled to a different rule as regards his insurer from that proper between the original sufferer and his, we should have *680the anomaly of allowing only a partial recovery of so much of a cargo loss as remained unsatisfied after the average contributions had been paid, and of allowing full recovery of so much of a cargo loss as fell upon eargo contributors.

The libelant further argues that the doetrine does not apply equally, because it alleges that, if there has been a decrease in value, the underwriter in eases of general average contribution pays only that proportion which the contribution bears to the agreed value. So far as we know, the law has not yet been so settled, and it may never be. The only evidence that it is the custom is of a Belgian underwriter; we cannot find in this record that it is universal. If it is, the contract may indeed be controlled by it, but then only because of the custom. We cannot decide this case on that assumption. Moreover, a custom may operate unequally if it has been accepted, and need not be applied beyond its scope for purposes of consistency. On principle we cannot see why the rule should not apply equally to eases of average contributions as to partial losses. It is true that this would result in making the policy more than indemnity, but again that is the case in partial losses. There is no greater reason to deny it because the insured recovers more than his actual contribution than because he recovers more than the goods have been actually damaged. In either case the result arises from the contract, and, if the valuation be honest, violates the principle of indemnity no more than is essential if the parties are to be allowed thus in advance to fix their rights.

In general, therefore, it appears to us that, since coinsurance is .right in principle, and is admittedly the law as to partial losses, it should apply to general average contributions. We can find nothing to the contrary in the books, except British, etc., Co. v. Maldonado & Co., 182 F. 744 (C. C. A. 9), and an early New York case, Strong v. Firemen’s Ins. Co., 11 Johns. 323, in which the chief point discussed was as to the proper place at which the adjustment should be computed. The question of whether, when the amount was reached, the underwriter was liable in toto was assumed. Except for the law relating to hulls, we should be reasonably confident that the respondent was right, in spite of our unwillingness to differ with our brothers in the Ninth circuit.

The decisions touching hulls are, however, in conflict. In cases of partial hull losses the insured in England recovers in full, regardless of the valuation (Aitchison v. Lohre, L. R. 4 App. Cas. 755), and the law is the same here (International Nav. Co. v. Atl. Mutual Ins. Co., supra). Such eases come up usually when the ship has been repaired, and a repair bill always represents a sum of money fixed without regard to the ship’s value, except for the rule of thumb, in eases of old ships, that only two-thirds of the bill shall be allowed. Were the insured treated as a coinsurer in such cases, the value of the ship would have to be found, and then the proportion which the repairs bore to it. Such values are not ordinarily (or at least were not when the rule became settled) fixed by a market; they rest upon opinion. There was some practical justification, therefore, for adopting the easier rule of taking the repairs as the measure of the difference between the ship if staunch and as injured. However, it is interesting to observe that, in a case where repairs were not made, but the ship was actually sold (Pitman v. Universal Marine Insurance Co., L. R. 9 Q. B. D. 192), Lindley, J., held that the doctrine of coinsurance applied. In the Court of Appeal this point was not argued, was apparently abandoned by the underwriter, and the ease did not decide it. The ruling is nevertheless important, as showing how much considerations of convenience may have determined the law as it stands.

On the other hand, in case of general average contribution by the hull, the House of Lords in S. S. Balmoral Co. v. Martin, L. R. [1902] App. Cases, 511, held the insured to be a coinsurer, thus assimilating the law to that governing partial cargo losses. It would therefore seem that, if the ease arose of average contribution by a cargo, the English courts a fortiori would use the same method of reckoning, and indeed this is conceded to be the law and custom in England, France, Germany, Holland, and Japan. The Supreme Court of Massachusetts so held in respect of hull contributions in the early case of Clark v. United, etc., Co., 7 Mass. 365, 5 Am. Dec. 50. We, however, decided the contrary in International Nav. Co. v. Atl. Mut. Ins. Co., supra, and in Int. Nav. Co. v. Sea Ins. Co. (C. C. A.) 129 F. 13, and in this we are in accord with the law of New York (Providence, etc., S. S. Co. v. Phœnix Ins. Co., 89 N. Y. 559).

It would indeed be possible to give as a reason for the English distinction that, in the *681■case of general average contributions, resort was not necessary to actual valuations. We have already said that this seems to us inadequate, and our own law has at least the merit of consistency in treating all hull losses alike. However, it is true that, if practical considerations are all in all, something is to be said for the English rule. As much could not he said for one which made a similar distinction as to cargoes. There is no substantial inconvenience in .fixing partial cargo losses, and no more in fixing general average. Little would be gained in practice, and much, as it seems to us, lost in principle, by importing a doctrine applicable to hulls for especial reasons, when the companion situation as regards cargoes is the opposite; and we have no reason to suppose that Brown, J., would not have treated all cargo losses alike, as he treated hulls. We agree that complete consistency is impossible. The question is whether we shall leave partial cargo losses the unique exception, or group them in one class and hulls in another. The less arbitrary division, and that most consonant with principle, appears to us to be the second.

Decree reversed; libel dismissed.