The merchandise involved in this case consisted, as imported, of 126 bales of granulated cork, of average weight of 175 pounds per bale. It was imported from Portugal on or about November 28, 1962, and was assessed with duty at 3 cents per pound under paragraph 1511 of the Tariff Act of 1930, as granulated cork, weighing not over 6 pounds per cubic foot uncompressed. It is claimed to be dutiable under said paragraph at 1 cent per pound as other granulated cork.
The pertinent parts of said paragraph 1511 are as follows:
Par. 1511. * * * clean, refined, or purified, granulated or ground cork, weighing not over six pounds per cubic foot uncompressed, 3 cents per pound; all other ground, granulated, or regranulated cork, 1 cent per pound; * * *.
According to the testimony, the mechandise was purchased in Portugal from the importer’s subsidiary in the form of “granulated cork,” *84which, was produced as follows: Cork wood or cork bark was broken into %-inch particles after being first dried to a given moisture content, normally 5 per centum. Thereafter, there was a gradual reduction in size through cutting operations, and grain, dust, bark, and other undesirable portions were separated out. The separable granulated cork was placed in a large baling apparatus and put under very severe pressure so that the cork was compressed to a density of over 10 pounds per cubic foot. The bales as imported were held together and kept under compression with sticks, wire, and burlap.
It is not contended that the merchandise is dutiable with respect to its density as imported. Both sides agree that the statute requires that the equivalent density “uncompressed” must be found. Broadly speaking, this requires that the effect of the preimportation compression be undone or neutralized, and the weight of the merchandise, whether over or under 6 pounds per cubic foot, be determined with respect to this uncompressed equivalent. The parties agree, by implication at least, that a sample or samples must be decompressed and the weight of the sample or samples or a portion of known volume be then ascertained. They differ in the details of how this is to be done.
In limine, the question arises as to what the statutory standard, not over 6 pounds per cubic foot, applies to: An entire shipment, or some portion thereof? Counsel have not briefed this question. For purposes of this case, we assume and hold it is not the whole shipment, but each unit as packed for shipment, that is, each bale. Counsel have apparently assumed that the bales are uniform and that whatever test weight is found to be correct shall apply to the entire shipment. However, if some bales are over and some under the 6-pound breaking point, only those under would be dutiable at the higher rate, unless the importer failed to differentiate or segregate, in which event, the higher rate might apply to the whole by virtue of section 508 of the Tariff Act of 1930, as amended.
We turn now to the different methods of measuring the density of the cork applied by the parties to the imported merchandise.
Apparently no sample was withheld from immediate delivery. The official papers, plaintiffs’ exhibit 7, and defendant’s exhibit B establish that 5 months after importation the appraiser notified the importer to submit a sample of 10 pounds of granulated cork from No. 1 bale. Evidence taken at the trial proves that the importer cut a piece out of the inside of a bale, a 30-pound section, with a saw, and delivered it to the examiner. The customs chemist, Mr. Graves, received this, finding it, as he testified, still -under compression. He said that the Customs Laboratory had no “standard method” for determining weight per cubic foot of granulated cork, but his office had received information from Armstrong Cork Co. and United Cork Co. The *85nature of this information was not admitted in evidence. Mr. Graves stated that, after removing the baling wire, burlap, and wood from the sample, he crumbled the cork particles with his hands and let them stand overnight. He then passed them through a sieve 0.157 of an inch in aperture. He poured “the sample” into a cylinder of glass, which held about 1/7 of a cubic foot. Apparently, he did this only once, discarding whatever of “the sample” the cylinder did not hold. He leveled off the top, and, of course, compared the weight of the cork contained in the cylinder with its known volume. On June 12, 1963, he reported that the sample weighed 4.8 pounds per cubic foot uncompressed.
Plaintiffs’ witness made tests of four bales of the cork, using a 6%-cubic-foot container or drum. Each bale was broken into small pieces and the pieces were forced through a screen placed over the mouth of the drum, so that the granules would separate and loosen. The drum was shaken to be sure it was filled and the excess scraped off the top. The drum was then weighed full and empty. From these tests, the witness obtained the weight of cork in the four bales. Four tests were made of each bale, but one test of bale No. 2 was insufficient because there was not enough cork to fill the drum. A tabulation of the results shows weights from 6.3 to 7.2 pounds per cubic foot, the average being 6.9 pounds per cubic foot.
It is well settled that the methods of weighing, measuring, and testing merchandise used by customs officers and the results obtained are presumed to be correct. United States v. Gage Bros., 1 Ct. Cust. Appls. 439, T.D. 31503; United States v. Lozano, Son & Co., 6 Ct. Cust. Appls. 281, T.D. 35506; Draper & Co., Inc. v. United States, 28 Cust. Ct. 136, C.D. 1400. However, this presumption may be rebutted by showing that such methods or results are erroneous. Sears, Roebuck & Co. v. United States, 3 Ct. Cust. Appls. 447, T.D. 33035; Gertzen & Co. v. United States, 12 Ct. Cust. Appls. 499, T.D. 40697; Pastene & Co., Inc. v. United States, 34 Cust. Ct. 52, C.D. 1677. Furthermore, the presumption does not have evidentiary value and may not be weighed against relevant and material proof offered by the plaintiffs. If a prima facie case is made out, the presumption is destroyed, and the Government has the burden of going forward with the evidence. United States v. Edson Keith & Co., 5 Ct. Cust. Appls. 82, T.D. 34128; Hawley & Letzerich et al. v. United States, 19 CCPA 47, T.D. 44893; United States v. Magnus, Mabee & Reynard, Inc., 39 CCPA 1, C.A.D. 455; James Bute Company v. United States, 33 Cust. Ct. 130, C.D. 1644.
In the instant case, plaintiffs have submitted evidence of tests which they applied to the merchandise and which gave a result different *86from that obtained by the Government. This evidence is sufficient to establish a prima facie case and the Government has, in fact, gone forward with the proof. It remains for the court to weigh all the evidence and determine which method and which result are the most accurate.
The various cases involving fluorspar provide a guide as to the kind of analysis and the quality of proof required where a tariff provision includes a “breaking point,” which requires extreme accuracy of measurement of the merchandise. Here, a rate 3 times as high applies to cork, weighing not over 6 pounds per cubic foot uncompressed, than that applied to cork weighing 6 pounds or over. In the fluorspar cases, the different rates applied to fluorspar, containing more than 97 percent calcium fluoride, and to fluorspar, containing not more than 97 percent calcium fluoride.
In the first case, Wm. A. Bird v. United States, 63 Treas. Dec. 980, T.D. 46437, it was established that there was a method of analyzing fluorspar adopted by the United States Bureau of Standards and that that method had been used by the commercial chemists but not by the chemists who testified for the Government. It further appeared that the samples used by the Government had not been taken in accordance with the regulations. The court held (p. 985):
* * * We are of the opinion that the testimony of the Government chemist with respect to his analysis of improperly identified samples by some method other than the one recommended by the Bureau of Standards is not sufficiently competent evidence to rebut the accuracy of analyses made by plaintiff’s witnesses, commercial chemists, based upon properly identified samples in accordance with the method approved by the Bureau of Standards. * * *
In the next case, T. H. Gonzalez v. United States, 40 Cust. Ct. 9, C.D. 1949, Government chemists had analyzed representative samples of fluorspar by the Bureau of Standards method, slightly modified. Three commercial laboratories, each using a different method of analysis, found different percentages of calcium fluoride in the merchandise. The Bureau of Standards method was not discredited or proved inaccurate. The Bidtel method, one of the other methods, was recognized to some extent by all the witnesses but had been discarded by many laboratories, including the Customs Laboratory. The court held that the analyses made by this method had not been shown to be more accurate than the determinations made by the Customs Laboratory and the protest was overruled.
In Ozark-Mahoning Company v. United States, 41 Cust. Ct. 16, C.D. 2015, representative samples of fluorspar had been analyzed by different chemists by different methods with varying results. The court found that both the Bidtel method' and modifications thereof and the Bureau of Standards method and modifications thereof were *87in use by analytical chemists and were accepted by other chemists in industry and that the evidence did not establish which method was per se the most accurate. It did find, however, that the results obtained by the Government chemist in analyzing the same sample by the same method varied by 0.11 percent in one instance, and by 0.21 percent, 0.24 percent, and 0.25 percent in other instances. The results obtained by each of the other chemists varied by no more than 0.05 percent and checked with the results obtained by the others within 0.25 percent. Their results were held to be the most accurate.
The most recent case was a rehearing of the Gonzalez case, supra. T. H. Gonzalez v. United States, 53 Cust. Ct. 149, C.D. 2487. The record established that the chief chemist of the Baltimore Customs Laboratory had investigated all available methods of analyzing fluor-spar and, with the use of synthetic samples, had found that the Bureau of Standards method was the only method that would give accurate results on all grades of fluorspar. An extensive series of tests by Government chemists and analyses by other chemists by this method showed a calcium fluoride content of less than 97 percent. The method used and the results obtained by the Government chemists were held to be correct.
These cases indicate that the final determination in situations where the merchandise approaches the borderline set by the tariff act depends upon the accuracy of the methods used and their application by the chemists who performed the tests. One criterion is whether the test has been established by an appropriate Government agency or is recognized by commercial laboratories or by the trade. Another is whether the results obtained check with a standard or with each other.
In the instant case, the Government chemist did not follow any procedure prescribed by regulation or in an official manual, and the Customs Laboratory did not have a standard method for determining the weight per cubic foot of granulated cork uncompressed. Use of an officially established method would present a different question with which this opinion is not concerned. The procedure used seems to have been invented ad hoe for the particular case, apparently on the basis of information about commercial tests which had been supplied by certain cork companies. No proof was offered as to what these tests were or to show whether they were used in circumstances that required the extreme accuracy of measurement necessary for the application of paragraph 1511.
The test as applied by Mr. Graves made use of but a small part of the sample delivered by the plaintiffs to customs officials, only about 11 ounces, if this court’s arithmetic is correct. This size sample thus governed the classification of 22,050 pounds of cork, more or less, or *88over 10 short tons. For such a result to be drawn from so small a sample, extreme niceness of weighing and measurement is required. The slightest error would be fantastically multiplied in the final result, yet an element of great uncertainty derives from the fact that the granulated particles were of differing sizes, as the procedure with the sieve established a maximum size, but no minimum. Furthermore, there is no recital of any precaution against the highly probable situation that the contents of different parts of the same bale might not be uniform. In addition, it seems at least possible that the manipulation of the sample and the screening of the particles, as described, might overadjust for compression; that is, the sample after these operations might be made to be of less average density than the entire bale before it was compressed in Portugal. Since the chemist made but one test, although a 30-pound sample of the merchandise was submitted by the importer, there is no way in which the court may determine whether his results were accurate. The record as presented is replete with reasons for doubting the intrinsic reasonableness and adequacy of the test and the accuracy of the results reached. In fact, about all that the Government’s brief argues in support of it is that the particles were allowed to stand overnight, during which time it is inferred that the particles became decompressed.
The tests applied by the plaintiffs resembled in method that of the defendant with the important distinction that the sample weighed and measured was about 1,000 times larger, nearly the whole of four bales. In the absence of other and perhaps more scientific methods of sampling and testing, the size of the sample inspires confidence that many conceivable causes of error were guarded against. Certainly, the use of the contents of the entire bale eliminated, or at least reduced significantly, the likelihood of error due to any lack of uniformity within the bale itself. Since 15 tests were made, it is possible to check the accuracy of the tests against each other. While there are variances, the largest within a single bale is %0 °f a pound in bale No. 1, and none of the tests showed a weight less than 6.3 pounds per cubic foot. As the record stands, these tests are more likely to be accurate and representative of the density of the entire shipment than the single one made by the Government chemist. General Aniline Works, Inc. v. United States, 6 Cust. Ct. 511, Abstract 45218; Philipp Bros., Inc. v. United States, 30 Cust. Ct. 216, C.D. 1523.
For the reasons stated, we hold that the merchandise in this case consists of granulated cork, weighing over 6 pounds per cubic foot uncompressed, properly dutiable at 1 cent per pound under paragraph 1511 of the Tariff Act of 1930, as other granulated cork. The protest is sustained and judgment will be rendered for the plaintiffs.