delivered the opinion of the court:
This appeal raises the issue as to the quantity of beer actually imported by the appellants at the port of New York and assessed for duty by the collector of customs at 23 cents per gallon. The importers, by proper protest filed in due time, claimed that they should have been required to pay duties only on the total number of gallons of beer which actually arrived and that the collector of customs ex*407acted from them duties on a quantity which was in excess of that really imported.
The Board of General Appraisers overruled the protest and the importers appealed.
It appears from the record that the merchandise was imported from Germany and Austria-Hungary in half barrels, which, before being put into use or whenever repaired, were officially gauged and marked with the official capacity in liters and sometimes in tenths of a liter. In Germany it seems that account is not generally taken of fractions of a liter, but in Austria-Hungary the practice prevails of marking on the cask the gauged capacity to the tenth of a liter, although some of the Provinces have found it convenient to ignore all fractions below five-tenths. As a rule, however, the brewers of the Austro-Hungarian Empire report in their consular invoices only the number of full liters in each package and make no mention at all of fractional liters. Every year in Germany and every second year in Austria-PIimgary all beer packages are regauged and after removal óf the old marks-rebranded on the head with the figures indicative of the capacity last found.
The beer which is here the subject of controversy on arrival at destination was discharged from the vessel on which it came at the Hoboken docks, port of New York, The importation was unladen under the direction of a foreman who represented the carrier, and whose duty it was not only to make delivery of the importation, but also to segregate all short, light, leaky, or damaged packages. The casks, which were apparently full and apparently in good condition, were after landing stood on end in rows, and the gauger, passing between the rows, noted in his dock book the name of the importer, the number and character of the entry, the number of each package, and its branded capacity in liters and in tenths of liters, if it was so marked. The gauger was followed by a stamper, who attached a customs stamp to each half barrel, tested it for shortage by tapping, and noted his estimate of any deficiency which might be found. The casks offered for delivery to the importers were also' inspected and tested by the importers’ truckman, who was personally responsible for all damaged and imperfect casks removed by him from the dock before adjustment. All short, light, leaky, or damaged containers, whether found by the steamship foreman, the stamper, or the importers’ truckman, were reported to the gauger, and he, after testing for himself the reported packages, noted in his dock book the quantity of beer lacking to fill the cask.
Under these circumstances, and as there was no practical way of measuring the beer in the barrels as they lay on the dock, the collector of customs, except in the case of packages shown by tapping to be deficient, took the branded capacity of the containers as proof of the quantity of beer imported, unless a greater quantity was in*408voiced, in which case, due probably to the fact that fractions of a liter were ignored generally in Germany and not infrequently in Austria, he accepted the invoice statement as correct. As already stated, the amount of beer in containers which were not filled or which were deficient was determined by experts, who carefully sounded the casks and estimated approximately the shortage.
The appellants contended before the board and contend here that the quantity of beer found by the collector to be imported was in excess of that which actually came into the country and that the half barrels contained on the average 13.33 instead of 15.56 gallons each, the quantity on which duty was collected. In support of this contention the inqporters claim that the evidence adduced by them on the hearing established that they had bottled 2,040 half barrels out of a total of 30,068 imported and that, as shown by the bottling, the 2,040 barrels contained 27,201 gallons, or an average of 13.33 gallons per barrel. In bottling the beer the importers used 270,860 bottles, of which number 10,220 were large-sized bottles and 260,640 were small-sized bottles. The importers determined the capacity of these bottles by filling 30 or 40 of them with water as nearly as possible to the same extent to which they would be filled with beer in commercial bottling — that is to say, to within about l-£ inches from the. top— and then measured the water by means of a United States standard gallon measure. Using this method of testing capacity, importers’ witness, Laxer, foitnd that it took about 5£ of the larger size and 10^ of the smaller size bottles to contain a gallon. The beer before bottling was kept in cold storage for about four days, which treatment deadened the effervescence to such an extent as to permit of the withdrawal of the bung and the siphoning off into bottles. If the beer showed no tendency to foam on the withdrawal of the bung, a high bung was inserted, and then the beer was siphoned into bottles by means of five pieces of not qxfite quarter-inch hose, each piece of hose being inserted and started before another was introduced, in order to prevent overflowing by reason of displacement.
The measurement of liquids by bottling them presents so many opportunities for error that it does not commend itself to us as an accurate or reliable method of determining quantities in any case and much less in a case where its acceptance would result in negativing results obtained by means which did not offer so much chance for mistake. It is true that in the present issue the bottling was done with some care. Nevertheless, considering the large amount of beer involved, we think it apparent from the record that the precautions taken were not sufficient to secure even a fair approximation to a correct result. In the first place, out of 10,220 large-sized and 260,640 small-sized bottles used in bottling the beer only 30 or 40 were tested for capacity. This percentage, we think, was altogether too small to establish either that all the bottles of each size were of *409uniform interior dimensions or that the variation, if any, in their cubical content was negligible. The witness Laxer did testify that all the bottles of each class were of uniform size, but that can scarcely be accepted as the equivalent of saying that they were of uniform capacity, especially as it is a matter of common knowledge that bottles apparently of the same size are not always capable of containing the same quantity of liquid. If, however, Laxer really meant to say that the bottles were all of uniform capacity, then his assertion must be accepted with some reserve, considering that the capacity of the bottles was a matter concerning which he could have no definite information beyond that afforded by the bottling test itself. In fact, Laxer’s statement that it took about 5-J- bottles of the larger size and 10¿ of the smaller size to hold a gallon of water, taken in conjunction with his subsequent declaration when called as a witness for the Government that beer bottles were of two sizes, one of about 5-f bottles and the other of about lOf bottles to the gallon, is hardly compatible with the idea that the bottles were exactly of the same capacity. It may be, as suggested by appellants’ counsel, that the testimony of Mr. Laxer as to the capacity of beer bottles was not correctly transcribed, but, if so, no such error is disclosed by the record, and even if it were the fact would still remain that he was testifying at best to nothing more than the capacity of 30 or 40 bottles out of more than 270,000.
In the second place, and apart from the fact that the number of bottles selected for the test was insufficient to justify a safe conclusion as to the capacity of the total number of bottles used, the accuracy of the test itself was open to serious question. It is not pretended that the test extended any further than the ascertainment of the number of bottles and fractions of a bottle which were required to hold a gallon of water. Such a test furnished no sound' basis for the conclusion that each bottle held exactly the same quantity of liquid as every other bottle and -consequently the test was not only valueless for the purpose of proving that the bottles were of the same capacity, but it furnished no guaranty whatever that the bottles did not so vary in capacity as to make them unreliable measures of liquids. Moreover, instead of ascertaining how many bottles it took to hold a gallon, the capacity of each of the bottles used for the test might have been accurately determined, and that that course was not pursued takes on a special significance in view of the fact that the test which the witness Laxer chose to make permitted him to say nothing more than that it took about 5£ bottles of one size and 10J bottles of the other to hold a gallon. The fact that the tested bottles were filled with water as nearly as possible to the point to which they would be filled with beer, coupled with the circumstance that subsequently 270,860 bottles were filled with beer at cold-storage temperature presumably to about the same point that the tested bot-*410ties had been filled with water at an undetermined temperature, did not conduce to accuracy of results and was by itself sufficient to impair the value of the test as evidence that the half barrels were 2 gallons short of their branded capacity.
It is true that the capacity found by the- method employed by the importers may have been approximately the correct capacity of each of the bottles tested. It is equally true, however, that it does not appear that that approximation was' so close that serious error could not occur when it came to bottling a barrel, to say nothing of 30,000 barrels. A mistake of a quarter of a gill per bottle in testing the bottles and filling them with beer would by itself mean an error of more than a gallon per barrel without taking into account the necessary wastage in siphoning.
With no evidence before us other than the presumption of correctness attaching to the collector’s decision the bottling test might be accepted as some proof of quantity, or it might be given greater weight than some less certain means of measuring liquids. Here, however, in the face of the presumption of correctness attaching to the collector’s decision, the presumption that the barrels were correctly gauged and branded, the fact that the beer was bought abroad and sold in the United States on the faith of the branded capacity, and the presumption that the filled packages contained the quantity of beer indicated by their brands, we are asked to accept the bottling test as decisive of the quantity of beer imported and as final proof either that the barrels contained less than their branded capacity or that they were gauged and branded in excess of their capacity. That we are not prepared to do, inasmuch as it would mean in effect the rejection of a more accurate and the favoring of a less accurate means of liquid measurement. Moreover, since it is not apparent why the casks which were opened for bottling could not have been gauged and the contents thus ascertained by a method which did not leave so much opportunity for mistake as bottling, and since it is certain that the 2,040 barrels emptied by bottling could have been gauged and their capacity accurately determined, we are unwilling to accept the bottling test in lieu of the better and stronger evidence which might have been produced. Of course, if the casks were not filled the bottling test might be well regarded as better and more satisfactory evidence of quantity than the branded capacity. In that connection it may be remarked, however, that although 2,040. barrels were opened for bottling by the importers, no testimony whatever was offered to show that a single barrel was found actually deficient. Indeed, far from being deficient, the witness Laxer found them so well filled that it was impossible to insert more than one piece of not quite quarter-inch hose at a time without causing the packages to overflow, notwithstanding that the thickness of the stave at the bung allowed some space for displacement. The witness Tielenius did testify on behalf *411of the importers that barrels of beer were not filled to their full capacity “ because they would bust.” As that testimony was apparently not founded on any actual knowledge or experience of the witness, but rather on his theory and opinion of what would happen if beer barrels were filled to the limit, we are not inclined to give it the weight which might have been conceded to it under other conditions. Moreover, the theory and opinion of the witness was flatly contradicted by the testimony of the Government inspector, who visited all the breweries which made the beer and witnessed the process of filling the barrels. This witness testified that—
The whole barrel is filled with beer without froth, and the only vacant space left is the bunghole. A workman with a hammer quickly drives a bung into this vacant space, so that the bunged-up barrel is entirely full of beer.
Admittedly there were some of the barrels imported which were not filled, but as all of the barrels imported were inspected for shortage by the importer, the steamship company, the customs gauger, and the customs stamper, we must assume that no unfilled barrel escaped attention. So far as appears from the record none of the barrels which were found deficient were submitted to the bottling test or to any better method of measurement than that which was adopted by the collector. As to such barrels we must ássume that the estimate of shortage made by the gauger, sustained as it is by the collector’s decision, was correct.
The decision of the Board of General Appraisers is affirmed.