Opinion of the Court by
Mr. Justice Harlan, announced by Mr. Justice Clark.These cases involve the validity of railroad tariff provisions exonerating the appellee railroads from liability for stated percentages of damage to shell eggs shipped over their lines. The cases come to us by direct appeal1 from a judgment of a three-judge district court in Utah,2 which dismissed an action brought to set aside and enjoin an order of the Interstate Commerce Commission 3 approving such tariff provisions. We noted probable jurisdiction on October 14, 1954.4
Claims against the railroads for damage to egg shipments steadily and rapidly increased in the years following 1939, particularly on shipments to the eastern seaboard area.5 In 1950 the railroads, believing that because of the difficulties of proof they were being exposed to liability for damage for which they were not responsible, filed with the Commission proposed tariff provisions similar in form to those approved by the order under review. After an investigation and hearing,6 the Commission concluded *164that egg shipments ordinarily contained substantial amounts of damage for which the railroads were not responsible — namely, (a) damage existing prior to shipment, and (b) damage unavoidably arising in transit because of the inherently fragile nature of eggs. The average amount of such damage was found to be 3% for eggs packaged at railhead points and 5% for those packaged elsewhere. On the basis of this finding, the Commission, although rejecting the higher-percentage provisions proposed by the railroads,7 found reasonable— and hence authorized the railroads to include in their tariff schedules8 — the following tolerance provision :
“On eggs placed in packages at rail point of origin of the shipment, no claim shall be allowed where the physical damage to the eggs at destination does not exceed 3% of the contents of the packages containing damaged eggs. Where damage exceeds 3%, claims shall be allowed for all damage in excess of 3%, if investigation develops carrier liability.
“Exception. — Where bona fide certificates of Federal or State egg inspection agencies showing extent of physical damage to eggs determined at rail point of origin of the shipment immediately prior to tender for rail transportation indicate the actual shell damage to be other than 2%, the percentage of actual damage as shown on such certificates, plus 1% shall be used in lieu of 3% specified in this Section.”
An otherwise identical provision applicable to “eggs placed in packages at points other than the rail point of *165origin” was determined to be reasonable with a tolerance of 5%.
It is claimed that these tariff provisions violate §20 (11) of the Interstate Commerce Act, 24 Stat. 386, as amended, 49 U. S. C. § 20 (11), which provides that any common carrier subject to the Act receiving property for interstate transportation “shall be liable ... for any loss, damage, or injury to such property caused by it ... , and no contract, receipt, rule, regulation, or other limitation of any character whatsoever shall exempt such common carrier . . . from the liability hereby imposed . . . .”
The Commission and the court below (one judge dissenting) held that the tolerance provisions did not violate §20 (11) because the pre-shipment and unavoidably-caused damage represented by the tolerances was not damage “caused by” the railroads; hence the tolerance regulations, in providing a means for determining the extent of such damage, did not limit the railroads’ proper liability, but operated simply to eliminate from damage claims the damage for which the railroads were not liable.
The appellants attack the provisions on six principal grounds: (1) the Commission has no jurisdiction over damage claims and hence no power to prescribe regulations governing their disposition; (2) tolerances based on averages necessarily embrace a forbidden limitation of liability since, by definition, some shipments will contain less than the “average” damage, resulting in those cases in the carrier being relieved of its full liability; (3) the railroads are liable for in-transit damage even though “unavoidable” ; 9 (4) the averages found by the Commission are not supported by the evidence; (5) the approval of uniform nation-wide tolerances was unreasonable in light *166of the wide differences in the egg-damage experience of consignees located in different areas of the country;10 and (6) the conclusion that the tolerances do not limit liability is not supported by the Commission’s findings. Our agreement with this last contention makes it unnecessary for us to consider the other arguments, and we may assume, though we do not decide, that the tariff provisions are not invalid for any of the other reasons assigned.
The Commission’s justification of the tolerance regulations as not limiting liability rests upon two distinct propositions: (1) that there is present in every case of eggs at destination physical damage not “caused by” the railroads — and hence for which they are not liable under § 20 (11) — in the amount of the specified percentages; and (2) that the deduction of those percentages from damage claims operates merely to prevent liability for such damage from being improperly imposed on the railroads. We *167shall accept for purposes of discussion the validity of the first proposition. The infirmity we find in the Commission’s report is rather that the second proposition is simply assumed and is supported by no findings upon which we can say that the Commission’s conclusion was reasonably based. Such a conclusion being essential to remove the tolerance provisions from the prohibition of § 20 (11), the lack of findings necessary to justify that conclusion renders invalid the Commission’s order approving the tolerance regulations. See Florida v. United States, 282 U. S. 194, 215. Indeed, so far does the report fail to support the Commission’s conclusion that it tends affirmatively to support precisely the opposite conclusion— namely, that the tolerances do unlawfully limit liability. We know of no better way to illustrate the inadequacies of the report than by showing the manner in which the inferences raised by it and unanswered by the Commission would, if accepted, lead to that opposing conclusion.
In the first place, we are unable to discover in the report any showing that damage claims include — or should reasonably be deemed to include — the exempt damage which is to be deducted from them. At common law, proof that a case of eggs contained a specified amount of damage for which the carrier was not liable would afford no defense to a damage claim not shown to include that damage. To complete the defense, some showing that the damage claimed included the exempt damage would be required, such as evidence that all of the damage had been found and claimed. Similarly, to justify a regulation authorizing the deduction from damage claims of a tolerance representing exempt damage, some basis for inferring that damage claims ordinarily include such damage would seem required, such as a finding that the type of inspection upon which damage claims are based is adequate to reveal substantially all the damage present in a case.
*168Far from making such a finding, however, the Commission report indicates that damage claims normally do not include all the damage present in a case. As described by the Commission, the customary inspection at destination upon which damage claims are based is simply a visual examination of a layer of 36 eggs at a time, continuing only until a layer in the case with no visible damage is found, at which point none of the succeeding layers is even looked at.11 Inasmuch as “damage” of the sort represented by the tolerances includes even the most minor shell imperfections, the inference seems inevitable that much damage is overlooked. This inference is strengthened by the Commission’s statement that much of the damage normally present in a case could be found only “by candling and clicking, and could not be detected by the kind of inspection performed at” destination.12 And that damage claims do not in practice include all damage is further indicated by the Department of' Agriculture studies cited by the Commission in which the damage overlooked by customary inspections at destination was found to range between 3.6% and 7.3%.13 Indeed, if the inferences suggested by these latter studies be accepted, it *169would appear that there is ordinarily overlooked in the destination inspection — and hence not included in damage claims — physical damage nearly equal to or greater than the exempt damage represented by the tolerances. If that be true, a further reduction of the claim by the full amount of the tolerance would necessarily operate to “limit” liability by approximately the full amount of the tolerance.
Nor is it an answer to this that the consignee is entitled to make a more thorough inspection than the prevailing practice entails. By in effect requiring a consignee to prove that his damage claim does not include the exempt damage, the tolerances would impose on the consignee the burden of disproving a defense which at common law it would be the carrier’s burden to establish. Whether or not the Commission has power so to alter the burden of proof, there is nothing to indicate that it had any intention of doing so. The Commission seems to have believed that under the prevailing commercial practices the carriers were being exposed to an improper liability and that the tolerances, applied in the same commercial setting, would do no more than remedy that situation. It would pervert the Commission’s purpose to deal realistically with a commercial problem now to seek to justify the tolerances on the ground that it is technically possible for consignees, by departing from the normal commercial practices, to avoid the limitation of liability caused by the tolerances. Especially is this true in the absence of any suggestion in the Commission’s report that such a complete inspection by consignees would be commercially feasible.14
*170Another inadequacy of the Commission’s report arises from its failure to distinguish between different kinds of physical damage in the application of the tolerances. The most favorable evidence supporting the Commission’s conclusions as to the extent of the damage not “caused by” the railroads was a Department of Agriculture study showing that at destination the average case of eggs contains 4.8% “checked” or “stained” eggs and 0.3% broken eggs.15 Presumably, therefore, the tolerances approved by the Commission represent physical damage in the same proportions — that is, primarily checked and stained eggs with only a nominal percentage of broken eggs. Checked and stained eggs are salable at a reduced price and therefore, unlike broken eggs, represent only a partial loss. The tariff provisions, however, do not differentiate between these types of damage and apparently authorize the deduction of the full tolerance without regard to the nature of the damage claimed. But to permit the offsetting of checked and stained eggs, representing only a partial loss, against broken eggs, representing a total loss, would seem necessarily to limit the railroads’ proper liability under § 20 (11).
A third major respect in which the Commission’s report falls short of establishing that the tolerances will not limit *171liability results from its failure to consider the relationship between the physical damage represented by the tolerances and the legal loss for which damage claims are asserted. We have thus far assumed that “physical damage” could properly be equated with “loss” and have shown that, even on that assumption, the Commission has not shown the existence of a relationship between the damage represented by the tolerances and that included in damage claims which would justify the deduction authorized by the tariff provisions. In fact, however, it would appear that damage claims are based not on physical damage to individual eggs but rather on the loss of commercial accceptability of a case of eggs as a whole. As the Commission observed, the “commercially sound” case of eggs invariably contains some damage,16 and it is apparently only when the damage is so great as to make a case commercially unacceptable for its grade that it loses value. When a commercially unsound case is received, the consignee, rather than suffer the presumably greater loss that would result from selling it at a lower grade or as salvage, ordinarily “reconditions” the case by replacing the visibly damaged eggs and, if necessary, the packaging materials. From the character of this process as described by the Commission,17 it is apparent that the purpose is simply to make the case commercially sound and not to discover and remove all the damage present in the case. The damage claim, if liability is asserted against the carrier, then consists simply of the cost of reconditioning the case — that is, the labor and material costs plus the loss on the damaged eggs removed.18
*172Viewed in the above terms, the probable effect of the tolerances on liability becomes even more apparent. Inasmuch as the highest grade specifications prescribed by the Department of Agriculture permit physical damage of 5%,19 a case of eggs received at destination with only the tolerance damage of 3% or 5% would presumably be considered “commercially sound” and be salable at the full price. Thus it would seem that presence of the tolerance damage by itself causes the consignee no loss and affords no basis for a damage claim, and that it is only when there is additional damage, making it necessary to recondition the case to make it commercially sound, that there is a loss for which a claim may be asserted. And if the additional damage is caused by the railroad, it necessarily follows that the cost of the reconditioning made necessary only because of that damage is a loss “caused by” the railroad within § 20 (11) ,20 Any reduction of the damage claim in such circumstances would thus relieve the carrier of its proper liability.
The Commission’s report thus leaves us not merely with uncertainty as to the impact of the tolerances, cf. United States v. Chicago, M., St. P. & P. R. Co., 294 U. S. *173499, 504-505, 510-511, but with the strong impression that they are likely to operate in a forbidden manner. The report contains no answers to the problems we have raised, if indeed the Commission considered them at all. Since the report falls far short of establishing that the tolerances will not operate to limit carrier liability in violation of § 20 (11), the order of the Commission approving the tolerances must be set aside, and the judgment below
Reversed.
28 U. S. C. §§ 1253, 2325. The Secretary of Agriculture’s standing in these proceedings derives from 52 Stat. 36, 7 U. S. C. § 1291 (a) and (b).
Utah Poultry & Farmers Cooperative v. United States, 119 F. Supp. 846.
2841. C. C. 377.
348 U. S. 807.
The Commission found that on eggs shipped to New York the average claim per car increased more than 1800% from 1940 to 1947, 284 I. C. C., at 386, 390, and that the total claims paid in 1947 exceeded 50% of the gross revenue on such shipments, id., at 387.
36 Stat. 552, as amended, 49 U. S. C. § 15 (7).
The tariff provisions as originally filed established 4% and 6% tolerances as opposed to the 3% and 5% tolerances found reasonable by the Commission. See 284 I. C. C., at 407-408.
Promptly after the Commission's order, the railroads refiled the tariff provisions with the approved percentages. They were permitted to go into effect on May 2,1952.
It is conceded that §20 (11) codifies the common-law rule making a carrier liable, without proof of negligence, for all damage to the goods transported by it, unless it affirmatively shows that the damage was occasioned by the shipper, acts of God, the public enemy, *166public authority, or the inherent vice or nature of the commodity. See, e. g., Bills of Lading, 52 I. C. C. 671, 679; Chesapeake & O. R. Co. v. Thompson Mfg. Co., 270 U. S. 416, 421-422; Adams Express Co. v. Croninger, 226 U. S. 491, 506-507, 509. The “unavoidable” damage to eggs is claimed to be within the “inherent vice” exception. The appellants contend, however, that the exception does not include damage induced by external force, however slight or unavoidable, but is limited to loss or damage arising solely from the nature of the property without the intervention of human factors, such as loss from decay, fermentation, evaporation or natural shrinkage. See, e. g., Austin v. Seaboard R. Co., 188 F. 2d 239, 240-241 (C. A. 5th Cir.) ; Akerly v. Railway Exp. Agency, 96 N. H. 396, 400-402, 77 A. 2d 856, 860-861; Jackson & Perkins Co. v. Mushroom Transp. Co., 351 Pa. 583, 590-591, 41 A. 2d 635, 639; Watson Bros. Transp. Co. v. Domenico, 118 Colo. 133, 135-137, 194 P. 2d 323, 325.
The Commission found that the increase of damage claims to “substantial amounts” had occurred principally in connection with shipments to certain points in the eastern seaboard territory — particularly New York, Baltimore, Philadelphia, Boston and Newark—and that “Generally there has been no increase in the damage claims on shell eggs moved by rail to other territories.” 284 I. C. C., at 385-386, 392.
2841. C. C., at 395.
2841. C. C., at 393-394. In “candling,” each egg is placed before a light to disclose defects not otherwise detectable; in “clicking,” two eggs are knocked together, the sound revealing shell imperfections.
284 I. C. C., at 396. In answer to the implications of these un-controverted studies, the Commission simply said: “The record does not indicate the handling received by those particular shipments after they were delivered to the consignees. It does indicate, however, based on the number and amount of claims, that the inspection of shipments of eggs arriving at New York is being performed with a view to detecting all damaged eggs. Claims in amounts exceeding 50 percent of the revenue on the entire egg traffic to New York in 1947 do not appear to be based on a cursory inspection.” Ibid. But this speculative reasoning hardly overcomes the Commission’s own explicit finding of the inadequacy of the destination inspections to discover all the damage.
To the contrary, the report notes that, despite the efforts of packers by careful inspection to eliminate all damage, “absolute perfection is commercially impossible” and a substantial amount is overlooked, 284 I. C. C., at 393, and the difficulties would seem even *170greater under the more adverse conditions prevailing at destination. Nor do the Department of Agriculture test studies upon which the tolerances were based indicate the feasibility of an inspection adequate to uncover all damage. The usual Department of Agriculture grading inspections involve an examination only of 100 eggs from each of 15 cases out of a carload, 284 I. C. C., at 393, and, while the Commission does not clearly say so, apparently the test studies utilized a similar spot-check inspection, from which the total amount of damage was simply inferred. Damage claims, on the other hand, can be asserted only for the damage that is actually found.
284 I. C. C., at 393-394. “Checked” eggs have slight cracks but the membrane is unbroken; “stained” eggs are sound eggs which have been stained by leakage from other eggs, /d., at 384, 385.
2841. C. C., at 393.
284 I. C. C., at 395. The eggs are simply transferred from the damaged case to a new case a whole layer (36 eggs) at a time, in the course of which any obviously damaged eggs are removed and replaced.
2841. C. C., at 385, 399.
284 I. C. C., at 401. Of this 5% (for Grade A or AA eggs), no more than 0.5% may consist of broken eggs. As appears above, however, the proportion of broken eggs represented by the tolerances is well within this limit.
It seems immaterial that in a given case the consignee might remove more damage than the minimum necessary to make the case commercially sound. The cost of the extensive inspection that would be necessary to assure that precisely the right amount of damage is removed — a cost the consignee would presumably be entitled to recover — would defeat the mitigation of damages which is the very purpose of the reconditioning. The carrier can ask no more than that the consignee act reasonably in mitigating damages, and there is no suggestion that the commercially accepted method of reconditioning unsound cases is unreasonable.