MEMORANDUM AND ORDER
Plaintiffs move for judgment on the administrative record of Electrolytic Manganese Dioxide from Ireland; Final Determination of No Sales at Less Than Fair Value, 54 Fed.Reg. 8,776 (1989) (Final Determination ) and ask the court to remand this matter to the International Trade Administration (ITA or Commerce) for a new determination of whether electrolytic manganese dioxide (EMD) from Ireland was or was likely to be sold in the United States at less than fair value (LTFV). Plaintiffs ask that the court direct ITA to extend its period of investigation (POI) to capture certain United States sales made by Mitsui Denman (Ireland) Ltd. (MDI). Plaintiffs also ask that ITA be required to use a standard for determining whether sales at LTFV are likely which does not require evidence of an irrevocable offer of sale, and that ITA be required to conduct a broader investigation, including verification at the premises of all potential U.S. purchasers of EMD.
FACTS
EMD, manganese dioxide (Mn02) that has been refined in an electrolytic process, is an intermediate product used in the production of dry cell batteries. On May 31, 1988, Kerr-McGee Chemical Corporation and Chemetals, Inc. (plaintiffs), the principal U.S. producers of EMD for sale on the open market, filed an antidumping petition on behalf of the U.S. industry producing EMD, alleging that EMD imports from Japan, Ireland, and Greece were being, or were likely to be, sold in the U.S. at LTFV and that such imports materially injure, or threaten material injury to, the domestic industry. See Plaintiffs ' Petition for the Imposition of Antidumping Duties, Public Record Document (P.R. Doc.) 1.
On June 27, 1988, ITA initiated an anti-dumping investigation with respect to EMD from Ireland. Initiation of Antidumping Duty Investigation; Electrolytic Manganese Dioxide from Ireland, 53 Fed.Reg. 24,115 (1988). On the same day, it also initiated investigations with respect to EMD from Greece and Japan. Initiation of Antidumping Duty Investigation; Electrolytic Manganese Dioxide from Greece, 53 Fed.Reg. 24,114 (1988); Initiation of Antidumping Duty Investigation; Electrolytic Manganese Dioxide from Japan, 53 Fed.Reg. 24,116 (1988).
According to plaintiffs, ITA contacted them on August 1, 1988, to advise them that the Office of Investigations was too occupied to conduct the Irish EMD investigation, which would therefore be conducted by ITA’s Office of Compliance, “an office of the ITA that does not generally conduct investigations under section 731 of the [Tariff] Act [of 1930].” Memorandum in Support of Plaintiffs’ Motion for Judgment upon the Agency Record (P.Brief) at 5. The transfer of authority took place during the composition of ITA’s questionnaire.
In their comments on ITA’s draft questionnaire plaintiffs asserted, inter alia, that “most EMD purchasers observe an annual purchase cycle” and that “sales of EMD to U.S. battery producers are governed by long-term contracts of up to 3 years.” P.R.Doc. 10, at 350. Commerce attempted to address plaintiffs’ concern in designing its antidumping questionnaire. Commerce nevertheless decided to adhere to its standard practice and chose a six month POI, that is, December 1, 1987 through May 31,1988. P.R.Doc. 20, at 602.
MDI, the sole producer of EMD in Ireland, is 75% owned by MMS, a major Japanese EMD producer, 15% owned by Mitsui & Co., Ltd., Tokyo, 5% owned by Mitsui & Co., Ltd., London, and 5% owned by the Industrial Development Authority of Ireland. P.R.Doc. 76, at 755. Mitsui New York is Mitsui Tokyo’s United States trading company. Id. at 754.
*616MDI’s primary purchaser, and another corporation described here as the potential purchaser, comprise two of the three major purchasers of EMD in the United States. P.R.Doc. 14, at 283. Both purchasers also produce EMD. The primary purchaser produces it solely for its own use. Id. at 281. In late spring of 1987, a fire disabled the cell room of MDI’s primary purchaser’s EMD plant, forcing the purchaser to obtain all its EMD requirements from outside sources. P.R.Doc. 14, at 281.
Before a U.S. purchaser will use EMD of a particular kind from a particular supplier, the product must be “qualified”.
The initial two qualification steps involve (step # 1) the chemical and physical analysis of the material followed by (step # 2) a laboratory scale battery test. This would typically require providing the customer with a 1 to 2 kilogram (“kg”) sample. Step # 1 usually involves one to three months to complete although if the tests need to be repeated, the time can be significantly longer. In the second step several hundred simulated batteries (batteries not jacketed or packaged) are produced for testing purposes. The tests include those for tolerance and integrity. This latter step can take between six months to a year to complete.
Based on an acceptable product performance in the first two steps, qualification would proceed to (step # 3) a semi-line scale test. In this step the customer would be provided with a large sample size (estimated average of between 10-20 MTs [metric-tons]) in order to use the material in a small scale battery production run. The completed batteries produced in this run are usually stored for a period of six months (simulating the average 6 month shelf-life prior to sale) before they are subjected to a series of additional tests. Because of this storage period, step # 3 is the longest step in the qualification process, and typically averages between six months to a year.
The final step of qualification, the final line scale test (step # 4), would be the use by the potential customer of an even larger sample size of up to as much as 250 MTs for trial plant production. This requires stopping production and cleaning the line beforehand in order to run the product in the plants [sic] actual battery production line. The purpose of this last step, which is the shortest step to complete, is to check the processibility of the product and assure that no physical or mechanical problems arise in its expected operating environment. For these reasons, this final step needs to be done in the plant where production would physically occur.
P.R.Doc. 76, at 757-58.
Following issuance of the questionnaire and follow-up inquiries, counsel for MDI submitted a letter to Commerce which stated that:
(1) MDI had made no shipments of EMD to the United States during the investigation period;
(2) Neither MDI, nor its agents or affiliates[,] entered into contracts during or prior to the investigation period to make shipments of EMD produced by MDI to the United States during the investigation period;
(3) Neither MDI nor its agents or affiliates made any offers of sale during the investigation period of EMD produced by MDI;
(4) There are no open pre-existing contracts entered into prior to the investigation period which call for the shipment of EMD to the United States by MDI; and
(5) To the best of MDI’s knowledge and belief, no EMD sold by MDI to third countries was later resold to the United States.
P.R.Doc. 25, at 465-66.
By letter dated September 15, 1988, plaintiffs responded to MDI’s assertion and alleged that Bureau of Census data showed that there had been imports of EMD from Ireland since at least 1978 (when the bureau first began recording imports of EMD) until 1987, P.R.Doc. 27, at 476, and that the POI should be extended backward to begin in January of 1987 and thus include sales of EMD from Ireland reported as imports in January, March, and May of *6171987. Id. at 479, 482. As a reason justifying such a retroactive enlargement of the POI, plaintiffs alleged that MDI’s sales activity during the POI was “unusually depressed.” Id. at 481-82 (quoting Certain Iron Metal Castings from India; An-tidumping Final Determination of Sales at Less Than Fair Value, 46 Fed.Reg. 39,869, 39,870 (1981)).
By letter dated September 21, 1988, MDI replied that an extension of the POI was inappropriate according to ITA regulations and prior rulings. P.R.Doc. 32, at 523-29. MDI also stated, inter alia, that its primary U.S. purchaser had disqualified MDI EMD in early 1987 and that MDI’s product remains disqualified for use in the primary purchaser’s batteries. Id. at 520.
On October 17, 1988, ITA asked MDI to confirm information regarding its most recent sales to United States purchasers. P.R.Doc. 42, at 599. MDI responded to ITA’s request for information by letter dated October 24, 1988. P.R.Doc. 46, Confidential Record Document (C.R.Doc.) 13. MDI stated that its primary U.S. customer had “disqualified MDI’s EMD for use in the United States in December 1986, at which time there were no open sales agreements between [MDI’s primary customer] and MDI.” C.R.Doc. 13, at 251A. MDI had sent one shipment to its primary customer in early April of 1987, which it had requested for testing purposes. P.R.Doc. 46, at 248-49.1
The record also reflects, however, that at a meeting in late 1987 between representatives of Mitsui and MDI’s primary purchaser, the latter reported very positive results from testing of MDI’s new EMD product, and stated that more would be required for further testing. C.R.Doc. 24, at 517A-18A. MDI officials reported that in February, 1988, the primary purchaser expressed interest in obtaining a larger quantity of MDI’s new EMD product. This was followed by an inquiry to Mitsui NY in May, 1988, for a specific quantity. P.R.Doc. 76, at 764-65. ITA verified that Mitsui NY had received purchase orders from the primary purchaser in June, 1988, which the primary purchaser intended to use for step # 4 of qualification. Id. at 765.
Mitsui NY told ITA, however, that at Mitsui NY’s request, the primary purchaser canceled the entire order. Id. The parties dispute whether the entire order was canceled. The initial June request comprised two orders. C.R.Doc. 24, at 610A-13A. A change notice in the record appears to cancel only one of the two orders, id. at 614A, but apparently the purchase orders were not filled and can be considered effectively canceled.
Both of the “canceled” purchase orders refer to a “purchase agreement”, pursuant to which the orders may have been made. This purchase agreement did not appear in the original administrative record submitted to this court. At oral argument held on February 9, 1990, Commerce admitted that the agreement had been reviewed at verification. As directed by the court Commerce subsequently submitted the agreement to the court.2 Exhibit A to Defendant’s Supplemental Memorandum (D.Supp.Mem.).
Apart from MDI’s dealings with its primary purchaser, of interest are MDI’s contacts with the so-called potential purchaser.3 On October 25, 1988, plaintiffs submitted a letter to Commerce from an offi*618cial of this potential U.S. purchaser of MDI EMD. According to the letter, MDI wrote to the potential purchaser in November, 1987, describing its product as “PVS E.M. D.” Moreover, the letter stated that representatives of Mitsui and the potential purchaser met on April 21, 1988, during the POI, and that at the meeting the Mitsui representatives offered to sell PVS to the potential purchaser. P.R.Doc. 48, at 629.
In response, MDI’s counsel submitted a letter to ITA dated November 2, 1988. P.R.Doc. 54. MDI stated that contacts between it and the potential purchaser during the POI were based upon MDI’s efforts to “qualify” MDI’s EMD:
Like all battery manufacturers [the potential purchaser] requires that EMD from any potential supplier be subjected to a time-consuming and multi-step process of extensive analysis and testing before EMD is “qualified” for use in its production process. Only after a particular supplier’s EMD is so qualified, will a battery manufacturer enter into negotiations with the supplier regarding potential purchases. MDI’s contacts with [the potential purchaser] have never gone beyond the stage of attempting to achieve a “qualified” status for MDI’s EMD. Since MDI’s EMD was not qualified by [the potential purchaser], there has never been an occasion for negotiations or discussions between MDI and/or its representatives and [the potential purchaser] regarding pricing or other elements essential to a sales negotiation process. It would simply be premature to address such issues until [the potential purchaser] determines that MDI’s EMD is “qualified” and, to MDI’s knowledge, [the potential purchaser] has not yet made such a determination.
P.R.Doc. 54, at 651-52.
MDI also submitted a November 2, 1988 letter from the purchasing manager of the potential purchaser, stating that the potential purchaser
never purchases EMD from an outside supplier without first subjecting that supplier’s EMD to extensive analysis and testing to confirm that it is ‘qualified’ for use in [the potential purchaser’s] batteries. I believe this “qualification” process is standard operating procedure for all major battery manufacturers.
[The potential purchaser] has never qualified the EMD produced by [MDI] and has never made commercial purchases of EMD produced by MDI. (During the period 1978-1980, [the potential purchaser] purchased small quantities of the standard grade EMD then being produced by MDI for testing purposes. These are the only purchases [the potential purchaser] has ever made of MDI EMD.) Over the course of the last two years, we have had several preliminary discussions with representatives of MDI regarding the potential qualification of MDI’s alkaline grade EMD for use in [the potential purchaser’s] batteries. MDI has since provided [the potential purchaser] with several samples (free of charge) for testing purposes, but [the potential purchaser] has not yet qualified MDI’s EMD. Until [the potential purchaser] qualifies the MDI EMD, it might be premature for the parties to enter into specific sales negotiations regarding prices, quantities, etc. Indeed, there have been no such specific discussions between the parties. The most recent discussions between the parties regarding the potential qualification of MDI’s EMD occurred on April 21, 1988, when representatives of Mitsui Mining & Smelting Co., Ltd. (“MMS”) and representatives of Mitsui & Co. (USA), Inc. (“Mitsui”), a trading company which represents both MMS and MDI, met with me [and two other officers of the potential purchaser]_ The principal items discussed at this meeting were purchases by [the potential purchaser] of zinc powder produced by MMS (a product unrelated to EMD), and technical difficulties previously incurred by [the potential purchaser] in processing EMD purchased from MMS. (During the course of the discussion of these technical difficulties, we described our success in using EMD produced by another Japanese manufacturer.) In this connection, we requested the Mitsui representative to describe in general terms the various types of EMD *619produced by MMS and MDI and to compare those products to those of the other Japanese producer. The Mitsui representative, in describing alkaline grades of EMD produced by MMS and MDI (“TKY” and “PVS” grades, respectively), stated that in his view they were “as good as” the alkaline grade of EMD of the other Japanese producer, and urged us to complete the qualification of his material.
At the conclusion of this meeting, we said we would require additional tests of MDI’s products before its qualification could be considered for possible use in [the potential purchaser’s] batteries and we requested another sample for this purpose.
P.R.Doc. 55, at 659-60.
Following the preliminary determination 4 and verification in Ireland and at the Mitsui NY offices Commerce held a public hearing on January 23, 1989. An ex parte meeting between Commerce, plaintiffs’ counsel, and the purchasing manager of MDI’s potential purchaser was held on January 30, 1989. At this meeting, the purchasing manager discussed the April meeting between Mitsui representatives and representatives of the potential purchaser. The purchasing manager stated that MDI had offered to sell its EMD to the potential purchaser. P.R.Doc. 81, at 853. He also stated that
[s]inee we had already performed our lab tests for alkaline grade EMD, our next step was an LPR, or limited plant run. It was agreed at the April 21, 1988 meeting that 500 pounds of EMD would be shipped to us for this initial single batch plant trial. The next step of [our] qualification process would be a one-day run, then a week-to-week approval prior to final approval.
All of these trials may be carried on simultaneously, or almost simultaneously, dependent on initial results and ongoing tests, with final approval possible in five to six months.
Id. at 854-55. He stated further that “[completion of this qualification process could have been as early as fall 1988.” Id. at 855. In response to questions as to whether price terms of a potential purchase were discussed, he stated “[absolutely. I told the Mitsui representatives that unless they were price competitive we wouldn’t go through plant trials.” Id. at 856. It appears that the “competitive” price referred to would be the prevailing price at the time of a purchase order. Id. at 882-88.
As to whether the parties had discussed quantity, the purchasing manager replied
[y]es. I told them that we are going to buy from the outside, and I think that we talked about 1500 tons. It eventually ended up a [sic] approximately 3000-ton requirement because of our increased volume that took place in the last half of the year.
Id. at 858.
In response to the comments and testimony presented in the ex parte meeting of January, MDI submitted an affidavit from the Assistant Manager of the inorganic chemical department of Mitsui NY in which he denied that price had been discussed in the April meeting with the potential purchaser. P.R.Doc. 84, at 1014. He also stated that
[i]n marketing EMD to U.S. battery manufacturers ... Mitsui has never made an offer to sell at the “prevailing” price, the “market” price or at a “competitive” price or by making any other such general statement. When Mitsui makes an offer, it always specifies a precise dollar amount.
Id. at 1013.
A corporate officer of the potential purchaser stated, however, that
we qualify EMD on where we intend to buy that EMD if it passes our tests. Qualification is a costly process. It costs between $15- and $20,000. This cost is justified only if we intend to make purchases soon after the qualification process is complete.
*620P.R.Doc. 81, at 855. When asked if the potential purchaser ever conducted a limited plant run with MDI’s EMD, he stated that
[w]e did not. The EMD was sent to us on May 31, 1988, the date on which I understand this trade case was filed. We received the Irish EMD on June 9, 1988, and I was asked at that time by our management to evaluate the impact of the trade case on our decision to qualify Irish EMD.
It was my opinion that in light of the anticipated dumping margin we would be better served to qualify domestic materials for all battery applications. That will continue to be our policy if dumping duties of the magnitude preliminarily set for Mitsui Japan are also imposed on Mitsui Ireland. If they are not, we may very well revisit the Irish product.
Id. at 863-64.
To summarize, the record reflects purchase orders by the primary purchaser issued several days after the POI. The orders reference a prior purchase agreement and were canceled at the request of MDI’s U.S. trading company after the antidump-ing petition was filed. The record also reflects detailed negotiations with the potential purchaser, including an offer by MDI (of disputed specificity) to sell EMD to the potential purchaser. The petitions intervened and the offer was not accepted.
DISCUSSION
I.Standard of Review.
The court will uphold a determination of ITA unless it is “unsupported by substantial evidence on the record, or otherwise not in accordance with law.” 19 U.S.C. § 1516a(b)(l)(B) (1989). Furthermore, “the standard is not de novo.” Ceramica Regiomontana, S.A. v. United States, 10 CIT 399, 404, 636 F.Supp. 961, 965 (1986), aff'd, 810 F.2d 1137 (Fed.Cir. 1987). When reviewing a decision under the substantial evidence standard, “the Court will affirm the agency’s findings if they are supported in the record by such relevant evidence as a reasonable mind might accept as adequate to support a conclusion.” Alhambra Foundry Co., Ltd. v. United States, 12 CIT -, 685 F.Supp. 1252, 1255 (1988) (citations omitted).
II.Questions Presented.
Plaintiffs have presented three issues:
1. Did ITA abuse its discretion in refusing to extend the POI to cover the most recent sales in the United States by MDI?
2. Did ITA use an overly restrictive standard in determining that EMD from Ireland was not likely to be sold in the United States at LTFV?
3. Did ITA fail to conduct an adequate investigation?
The court will deal first with the issue of whether ITA erred in refusing to extend the POI backward to capture the last reported sales by MDI of EMD to U.S. purchasers. The question of whether ITA erred in refusing to extend the POI forward is closely linked to the question of whether Commerce used an overly restrictive standard for deciding whether United States sales are likely and will be discussed subsequently. The adequacy of the investigation will be discussed last, to the extent it is not covered in the earlier discussions.
III.ITA Did Not Abuse its Discretion in Refusing to Extend the POI Backward.
Congress has not mandated a length of time for antidumping investigations. The regulations in place during the investigation at issue stated
[и]pon publication of the notice of “Initiation of Antidumping Investigation”, the Secretary shall proceed promptly to obtain such information as may be necessary for preliminary and final determinations of sales at less than fair value. The Secretary normally will examine at least 60 percent of the dollar volume of exports to the United States from any country subject to an antidumping investigation. Ordinarily the Secretary will require a foreign manufacturer, producer, or exporter subject to the investigation to submit pricing information covering a period of at least 150 days prior to, *621and 30 days after, the first day of the month during which the petition was received in acceptable form. The Secretary may, however, require the submission of pricing information for such other period as he deems necessary and he may also require the submission of pricing information on a current basis during the course of an investigation. Where appropriate, cost information also will be required.
19 C.F.R. § 353.38(a) (1988).
Plaintiffs filed their petition on May 31, 1988. Therefore, in accordance with standard agency practice, ITA chose a POI which covered December 1, 1987 through May 31, 1988. Plaintiffs claim that the Secretary abused his discretion in refusing to extend the POI, claiming that the underlying theme of past agency practice is that the POI should be extended if “the six-month period of investigation by the Department does not accurately reflect [the exporter’s] sales activities in the United States.” P.Brief at 29 (quoting Petitioner’s Comment 1 in Forged Undercarriage Components from Italy: Final Determination of Sales at Not Less Than Fair Value, 49 Fed.Reg. 7,137, 7,138 (1984)). According to plaintiffs, “[t]he unifying theme of the ITA’s exercise of its discretion to extend a POI is that a POI longer than six months will be specified when the exporter under investigation is having a current effect on the market through sales activities during a conventional POI, and a longer POI is needed to measure fairly the less than fair value aspect of the sales or the benefit the sales are receiving from subsidization.” P.Brief at 29-30. Plaintiffs maintain that respondents conducted substantial marketing of MDI’s EMD during the POI and thereby affected price, even if no sales took place, and that therefore the POI should be extended to capture the last reported sales. Presuming that the last sales in the record were the 1987 sales to the primary potential customer and the secondary customer, this would require an eleven (11) month retroactive extension.
Commerce has extended the POI backward in past investigations if it found that unusually depressed sales during the POI would prevent ITA from conducting an adequate investigation. Certain Iron Metal Castings From India; Antidumping: Final Determination of Sales at Not Less Than Fair Value, 46 Fed.Reg. 39,869, 39,-870 (1981). In that investigation, the depression in sales lasted for a short period within the normal POI. In the investigation at bar, Commerce found that MDI made no sales in the U.S. from April, 1987 until February 13, 1989.5
The parties dispute the reason for the apparent lack of MDI sales. Plaintiffs claim that the phasing out of an outdated type of EMD and the switch to production of an improved version caused the period of sales depression. P.Brief at 35. Defendant, however, contends that the disqualification of MDI’s EMD by its primary purchaser and its nonqualification by others caused the drop. Defendant’s Memorandum in Opposition to Plaintiff’s Motion for Judgment upon the Agency Record (D.Brief) at 25-26. Whatever the reason, the sales depression extended for a considerable length of time.
Plaintiffs also allege that industry practice called for long term, often annual contracts, which render a six month POI inadequate. Commerce has extended the POI in situations where long-term contracts would skew the investigation, Final Determination of Sales at Less Than Fair Value, Certain Forged Steel Crankshafts from the United Kingdom, 52 Fed.Reg. 32,951, 32,951-52 (1987). In the investigation at bar, ITA determined that “no shipments of EMD from MDI were made during the POI which correspond to sales made prior to the period nor were shipments made during 1987 pursuant to long-term contracts with U.S. purchasers.” Final Determination, 54 Fed.Reg. at 8,777. See also C.R.Doc. 24, at 332A, 383A-84A. These conclusions are supported by the evidence of record.
*622Plaintiffs further contend that seasonality exists in the EMD market, with purchases “concentrated in the second half of the year anticipating the usual run up in the Christmas sales of batteries.” P.Brief at 32. ITA has extended the POI in the past to make up for distortion caused by seasonality. See Final Determination of Sales at Less Than Fair Value: Certain Fresh Cut Flowers from Colombia, 52 Fed.Reg. 6,842, 6,844 (1987). In the case at hand, Commerce found that
[t]he evidence for MDI, however, shows that when it supplied the U.S. market, its monthly shipment volume remained constant. This shipment stability is also evidenced by MDI’s related Japanese producer.
Final Determination, 54 Fed.Reg. at 8,777. See C.R.Doc. 24.
Evidence does exist in the record to confirm seasonal purchases by at least one purchaser. P.R.Doc. 81, at 842-43. The seasonality spoken of by an officer of this particular purchaser, however, indicates that the purchasing differential was not great. Id. at 843. Finally, as stated by Commerce, “[e]ven if seasonality were a factor and the POI were extended [backward] by an additional six months to capture a full year in [the] investigation (a sufficient period for eliminating distortions) no sales would be found within that extended period.” 54 Fed.Reg. at 8,777. The record supports this conclusion.
Plaintiffs allege that Commerce erred by limiting the criteria for extension of the POI to four circumstances previously found to warrant such extension.6 Plaintiffs complain that “[t]he ITA took the position for the first time in the proceedings below that if the facts do not fall within four general categories, it will not extend the POI.” P.Brief at 39. The court does not agree that the language of the final determination implies that Commerce intended to so limit the circumstances under which it would extend the POI, see Final Determination, 54 Fed.Reg. at 8,776-77, nor does Commerce claim to have done so. D.Brief at 29-30. Commerce merely listed various situations which have warranted extensions in the past and found that none of them applied in this investigation. It appears that Commerce did not, however, intend to present an exhaustive list.
Based on the totality of evidence in the record the court finds that it was not an abuse of discretion to refuse to extend the POI backward to nearly three times its normal length to capture earlier sales. Nothing in Commerce’s past practice should have indicated to plaintiffs that they had a right to expect ITA to so accommodate them. Certainly, ITA cannot be expected to extend the POI merely to achieve a result favorable to plaintiffs. Even if sales activity did affect prices, the extension backward which is sought is simply too great to be mandated by the facts of record.
IV. ITA’s Determination That Sales Were Not Likely During the POI Must Be Reconsidered.
As indicated previously, ITA found no sales by MDI from April, 1987 to February, 1989. Based on the chosen investigative period, ITA also found no likelihood of sales.7 Whether sales were likely and whether the POI should be extended forward to capture the sales are essentially the same question as to the primary purchaser. Extending the POI may or may not be technically necessary. Several reasons were given for the finding of no likely sales. The key reasons are all based on erroneous or unexplained standards as will be demonstrated.
*623A. ITA’s requirement of a finding of an irrevocable offer.
One of ITA’s key findings regarding no likelihood of sales is its finding of lack of an irrevocable offer of sale during the POI. In its final determination in this case, Commerce stated that
‘[m]ore than a speculative potential of future sales for export is necessary to meet the “likely to be sold” criterion of section 731 of the Act’ Certain Carbon Steel Products from Czechoslovakia, 50 FR 1912 (1985). The Department looks for evidence of a current offer, the acceptance of which is reasonably expected. See Dismissal of Antidumping Petitions on Certain Steel Products from Romania, 47 PR 5752 (1982). At the very least this requires evidence of an irrevocable offer to sell (Carbon Steel from Czechoslovakia).
Final Determination, 54 Fed.Reg. at 8,777. The irrevocable offer standard arose through an evolutionary process and has recently been codified in regulations not in effect at the time of the investigation at issue. According to the new regulation, “[a] ‘sale’ includes a contract to sell and a lease that is equivalent to a sale. A ‘likely sale’ means a person’s irrevocable offer to sell.” 54 Fed.Reg. 12,743, 12,771 (1989), to be codified at 19 C.F.R. § 353.2(t).
Plaintiffs contend that Commerce did not consistently require proof of irrevocable offers in the past. While conceding that ITA used this standard in Carbon Steel from Czechoslovakia and in Certain Carbon Steel Products From Poland; Initiation of Antidumping Duty Investigations, 50 Fed.Reg. 1,915 (1985), they claim that ITA did not apply this standard in Steel Products from Romania. In that investigation, Commerce found that sales were not likely because “no evidence ha[d] been provided of bona fide offers on merchandise for export to the United States.” 47 Fed.Reg. at 5,753. ITA concluded that “where no evidence has been furnished of a current offer the acceptance of which is reasonably expected, then no adequate allegation has been made of current or likely sales at less than fair value.” Id. at 5,752. These cases are inconclusive on the issue of whether a standard broader than the irrevocable offer standard was applied in the past. Nothing in ITA’s reasoning conflicts with an irrevocable offer standard. Plaintiffs have not cited any investigations in which ITA determined that sales were likely where only bona fide revocable offers existed. Therefore, the court does not find that the irrevocable offer standard in this investigation fails for lack of consistency.
More at issue is whether ITA’s standard fails for lack of precision or because it is overly restrictive. According to defendant, “[t]he ‘irrevocable offer’ standard applied by Commerce derives from basic principles of contract law.” D.Supp.Mem. at 4. Under traditional contract law, however, offers are revocable unless the offeree gives consideration to the offeror to keep the offer open. See, e.g., Farnsworth, Contracts § 3.23 (1982). In contrast, defendant states:
[although not bound by the rigorous requirements provided in the Uniform Commercial Code (UCC) for distinguishing offers which are irrevocable and those which are not binding on the offer- or Commerce considers the same basic elements in applying the standard: an offer which sets forth essential terms, such as price and quantity, coupled with the seller’s assurance that the offer will be held open for a certain period of time. Further, Commerce evaluates particular sales activities in light of the facts of a case. Accordingly, the relative importance of these factors will vary according to the conditions and practices in a particular industry, a particular course of dealing, or prior practice.
D.Supp.Mem. at 4-5 (footnotes omitted). Along these same lines, Commerce claims to consider “among other things, the following factors — a discernable price, an offer which is reasonably expected to be accepted within a foreseeable time, prior dealings, and industry practice,_” D.Bn 'f at 52.
First, to use the term “irrevocable” and then to state that it does not always mean *624“irrevocable” in the ordinary contract sense is misleading and does not alert affected parties to the true nature of ITA’s rationale in particular cases or to its general policies.
Second, Commerce defines “irrevocable” elsewhere as “imminent.”
The ‘irrevocable offer' standard ensures that sales which are imminent will be captured, without including sales activity which may or may not result in a sale. For example, applying the ‘irrevocable offer’ standard, Commerce would consider an offer drafted during the period of investigation which, although it contained essential terms and its completion was imminent (i.e., irrevocable), was not accepted during the period of investigation as a 'likely sale.’ Commerce would not consider as a ‘likely sale’ an offer to buy when there is no obligation to supply or accept the offer.
D.Supp.Mem. at 6-7. It is not clear to the court how the “irrevocable offer” standard ensures that imminent sales will be captured. Nor is it even clear that “irrevocable offers” are any more likely to be accepted than are revocable ones. Commerce appears to be claiming that an irrevocable offer to sell is more likely to lead to a sale than is a purchase order following detailed discussions, but it does not explain its reasoning for this assertion.
Third, it is not clear that ITA did take industry practice into account as it claims its practice dictates. When asked at oral argument if anyone in the EMD industry ever uses irrevocable offers, counsel for defendant answered that the investigation did not cover that point. Tr. at 94-95. If there was a reason for this, it was not explained in supplemental briefing. Therefore, the court fears that reliance on the so-called “irrevocable offer” standard might prematurely exclude a finding of likely sales in any industry which did not make use of whatever it is that ITA considers to be “irrevocable” offers. One would assume that Congress would not wish to automatically exclude such industries from consideration under a likelihood of sales standard.
Commerce has not sufficiently explained its rationale for determining what kind of offers are likely to be accepted and it may have applied an overly restrictive test in this case. Generally, it has defined its standard in unclear, misleading, and possibly contradictory ways. The court remands this investigation to Commerce to apply a likely sales test to the facts of this case which is clear and reasonable, and to explain it.
B. Commerce’s reliance on the fact that United States purchasers had not yet qualified MDI's new product.
Commerce also based its finding of no likely sales on the fact that prior to and during the POI U.S. purchasers of EMD had not “qualified” MDI’s new product. 54 Fed.Reg. at 8,777-78. According to defendant, “[u]nless such qualification was obtained, MDI could not sell its product to any U.S. purchaser.” D.Brief at 40. This conclusion, of course, conflicts with Commerce’s later assertion that sales for testing purposes only (part of the qualification process)8 are nonetheless sales for the purposes of the Act. Defendant states “[w]ith regard to sample merchandise which is intended to be used for testing purposes only, Commerce includes all United States sales of merchandise in its calculation of United States price for purposes of the less-than-fair-value comparison unless a party proves to Commerce’s satisfaction that a certain United States sale falls within a narrow exception which would warrant exclusion.” D.Supp.Mem. at 3 (footnote omitted). Presumably the narrow exception is the unrepresentative behavior exception. The record shows the sales for testing are part of the regular business practice of this industry. Neither defendant nor defendant-intervenor has cited to anything in the record to show that such sales for testing are so unrepresentative that they should not be considered for U.S. sales purposes.
Moreover, the record demonstrates that the primary purchaser had nearly complet*625ed the qualification process and that a potential purchaser had made considerable progress in this regard. Regarding the primary purchaser, the record reveals very specific negotiations for sales of merchandise to be used in the final stages of testing. See, supra, at 617. Furthermore, evidence in the record discussed, supra, at 619-620, indicates that prior to the anti-dumping petition, the potential purchaser had completed initial tests, had planned purchases and likely would have made them had the petition not been filed.
Given the record of substantial progress toward qualification and ITA’s view of the status of sales for testing, ITA’s finding of no likely sales is not supported by the fact of incomplete qualification in this case. Commerce must make a rational decision as to likely sales regarding both the primary and the potential purchasers applying the proper rules on sales and offers to sell for testing purposes, as well as whatever considerations relative to motivation are appropriate. See discussion, infra, at 627.
C. Defendant’s argument that no sales existed from April 1987 to February 1989.
As plaintiffs state, “[bjecause the Antidumping Petition was filed on May 31, 1988, the ITA elected a December 1, 1987, to May 31, 1988, POI.” Petitioner’s Reply Brief (P.Rep.Brief) at 23. Plaintiffs contend that “[h]ad the petition been filed one day later, the POI would have been January 1, 1988, to June 30, 1988, and would have covered MDI’s June ..., 1988, sales of EMD to” its primary customer. Id. Plaintiffs allege that it was an abuse of discretion not to extend the POI to capture these “sales.” Defendant now argues that the June “sales” were not sales because, inter alia, they were canceled. Apart from its bearing on the issue of extension of the POI forward, the existence of sales after the normal POI is important because, in its brief, defendant cites the lack of sales for an extended period before and after the POI as another reason why no sales were found to be likely. D.Brief at 43-44.9 The court does not believe it is clear, as defendant contends, that “[t]he record reflects that even if the period of investigation was extended forward to the last possible moment (until February, 1988 (sic)10) to capture any shipments made pursuant to the [primary purchaser’s] purchase orders, no sales or shipments were made by MDI.” D.Supp.Mem. at 4.
Commerce does not dispute that a sale takes place when a binding contract is formed. D.Supp.Mem. at 2. Setting aside for the moment arguments regarding subsequent cancellations, whether or not the June purchase orders represented sales depends upon whether MDI’s primary customer placed the orders properly pursuant to an earlier agreement to supply, in this case the 1988 purchase agreement (Agreement). If so, then a binding contract, specifying price and quantity, was formed.11 The purchase orders do refer to the Agreement. See C.R.Doc. 24, at 610A, 612A.12
The Agreement expressly refers to EMD produced by MDI. It appears, however, that the parties contemplated that MDI’s EMD would be used by the primary customer’s foreign plants. Moreover, the Agreement does not refer to the new type of EMD requested in the purchase orders. At page 1, however, the Agreement states that “[o]ther grades of Electrolytic Manganese Dioxide produced by MMS and MDI, in [the primary purchaser’s] determination, may be added to Material in this Agree*626ment by mutual written agreement by both parties.” Exhibit A to D.Supp.Mem. Such a written agreement, if it exists, is not in the record. The record does evidence a series of communications in advance of the purchase order and according to ITA’s Verification Report at least one of the purchase orders was canceled “[a]t the request of Mitsui NY.” P.R.Doc. 76, at 765. The cancellation appears in the record. C.R. Doc. 24, at 614A.
At oral argument, counsel for MDI stated that it did not wish to fill the orders at issue “because the price was too low.” Tr. at 116.13 Obviously, the remark was intended to support the proposition that MDI’s new EMD product had never been made part of the Agreement. The record itself is less than clear on this point. For example, the price listed on the purchase orders is exactly that of the EMD grade produced by MMS to which MDI told customers that its product was comparable and which MMS was bound to supply the primary purchaser. As Commerce states, “it is doubtful that a potential customer would even seriously consider making future purchases from a manufacturer who was not in a position to supply ... EMD at ‘a competitive price.’ ” D.Brief at 42. MDI’s contention that the price was “too low” is also undermined by its other arguments. The sales at issue were for qualification purposes, and MDI contends “that ‘sales’ made by an EMD producer such as MDI for the purpose of qualifying its product and thereby gaining access to the U.S. market would be priced lower than later sales.14 This is because until a purchaser ‘qualifies’ the seller’s product, the seller has no leverage with which to negotiate a regular price.” Mitsui Denman (Ireland) Ltd.’s Post-Hearing Brief at 4. [Footnote added]. Finally, MDI’s claim that the price was too low does not explain why Mitsui NY requested formal cancellation.
It Seems possible to conclude that after deciding, for whatever reason, not to fill the order, Mitsui NY would not have requested a cancellation unless it felt itself bound by the purchase order referencing the Agreement. It is for ITA to decide if some other rationale is better supported by the record. The record, thus far, indicates that ITA may have ignored the logic of the evidence and failed to consider the legal and practical implications of the two purchase orders made immediately after the filing of the antidumping petition. The record would support, at least, the conclusion that the parties reached a well-defined agreement in April or May, 1988, during the POI, when MDI and its primary customer discussed the subsequent orders. P.R.Doc. 76, at 765. Whether the parties actually made MDI’s new EMD part of the Agreement at that time is for ITA to decide.
At oral argument, Commerce did not dispute that the Agreement, if applicable to these purchase orders, constituted an irrevocable offer, under even a strict definition. Tr. at 89.15 Assuming arguendo that a classic irrevocable offer standard would be appropriate, it makes no difference that in order to determine whether an irrevocable offer was made during the POI, one must look to the purchase orders made several days after the POI and subsequent cancellations. Commerce consistently examines periods surrounding the POI in order to understand the significance of activity within the POI. It did so in this case, and based its conclusions on such examination.16 Furthermore, if a more flexible standard on likely sales is applied, whether or not MDI’s EMD was formally made part of the agreement, ITA might decide that sales were likely. Thus, ITA may have erred in basing its finding of no likelihood *627of sales upon the non-existence of sales after the POI, without considering the status of purchase orders, subsequently canceled, in connection with the Agreement. ITA also may have erred by not extending the POI to capture such sales. The actual facts are for ITA to decide in the first instance. The court is not convinced it has considered them to date.
Because apparently it did not consider the Agreement in connection with the purchase orders and did not decide whether sales had occurred, ITA did not address whether binding sales, later canceled, are “sales” for purposes of the Act. Defendant now claims that it does not regard the purchase orders as sales because, inter alia, “[d]uring verification, an official of Mitsui U.S.A. stated that the cancellation notice applied to both purchase orders, that both orders were indeed canceled, and that no EMD was shipped to [that customer].” D.Supp.Mem. at 3-4. If this is truly ITA’s rationale, it must be explained in terms of the likely sales standard of the antidump-ing law.
Commerce seems confused about its position as to canceled sales. At oral argument, in response to the court’s question as to how ITA regards canceled sales under section 731 of the Tariff Act of 1930, codified at 19 U.S.C. § 1673 (1988), defendant’s counsel stated that a canceled sale is still a sale. Tr. at 83. In its supplemental memorandum, however, defendant states that “if the contract is subsequently canceled, Commerce does not treat the transaction as a ‘sale’ for purposes of the antidumping law.” D.Supp.Mem. at 3 (citing 64k Dynamic Random Access Memory Component (64 Drams) from Japan, 51 Fed.Reg. 15,943, 15,947 (1986)).
The 64 Drams investigation referred to by ITA does not stand for the proposition for which it is cited. In that investigation, respondents complained that the Secretary erred in concluding that the date of shipment of the subject merchandise was the date of sale. Respondents contended that under Japanese law, offers are deemed accepted if not promptly rejected, and that therefore the date of the purchase orders should be considered the date of sales. Commerce stated that
[i]n general, the contract type of analysis set forth by Hitachi would be relevant in determining when a “sale” occurs for purposes of the antidumping laws. Here, however, the Department has determined that, in this particular industry, and during the time period investigated, neither party to a purchase order intended it to be a “binding agreement” or treated it as such. This was true for both the U.S. and home-market transactions. The department reached this conclusion based on the fact that during the time-period investigated, there were significant cancellations of [orders of the subject merchandise] by both parties, without any sanctions or penalties whatsoever, ... Such cancellations ... occurred even after shipment of the goods in question. Thus, the Department used the date of shipment for home-market sales since that was the earliest point in the transaction at which any sort of binding commitment may be inferred.
51 Fed.Reg. at 15,947. [Emphasis in original].
The passage quoted above clearly does not stand for the proposition that a sale, involving as it must parties who consider themselves contractually bound, is somehow undone for purposes of section 731 of the Act by a cancellation. Rather, it stands for the proposition that numerous cancellations meant, in one particular industry at one particular time, that neither party felt itself contractually bound and that therefore no sales had taken place prior to shipments.
The court cannot properly evaluate ITA’s determination unless ITA’s policy on canceled sales is explained. Furthermore, the question of motivation for the cancellation may or may not be relevant. ITA should explain if cancellations to avoid antidump-ing orders should be considered differently from other cancellations. Whether this was MDI’s motivation has not been decided.
*628D. Adequacy of ITA’s investigative technique.
The court agrees with Commerce that it is not required to conduct verifications at the premises of all potential purchasers of EMD. Congress intended that ITA have latitude in its verification procedures and that it not be required to comply with all requests for specific types of investigation. Monsanto Co. v. United States, 12 CIT-, 698 F.Supp. 275, 283 (1988). If ITA finds MDI’s information to be complete and its explanations sound, it may need no further information. If it makes a contrary finding, it has various options including use of best information otherwise available or purchaser verification. ITA has a range of reasonable options. The court will not order verification at all potential purchasers based on this record.
It also appears that a fourth purchase or potential purchase took place in December, 1988, P.R.Doc. 76, at 768, and that the shipment pursuant to this sale entered the U.S. in January, 1989. The shipment is documented in a Census Bureau report attached to plaintiffs' supplemental memorandum. This document is not part of the record. Since it appears, however, that this purchaser did not place its order until over six months after the close of the POI, the court cannot hold that ITA conducted an inadequate investigation for having failed to discover the sale or the shipment. Thus, ITA need not consider this sale on remand.
CONCLUSIONS
The court upholds ITA’s decision not to extend the POI retroactively, and cannot conclude that Commerce conducted an inadequate investigation by not conducting purchaser verification and in having failed to find that MDI shipped EMD to the U.S. in early 1989. The court, however, does not sustain ITA’s determination that there was no likelihood of sales of MDI’s EMD during the relevant period.
ITA shall define a standard for determining whether sales are likely which considers industry practice and which will capture imminent sales.
ITA shall explain its policy on canceled sales in terms of the governing statute and this case.
ITA shall determine whether U.S. sales existed or were likely. In doing so it shall consider the relationship, if any, between the primary purchaser’s orders and the preexisting Agreement, and it shall consider the activities discussed here of the primary purchaser, the potential purchaser and MDI.
As the court has asked ITA to define many concepts, ITA shall give the parties an opportunity to comment before issuing a final remand determination. To allow time for this additional step, the court will permit a 90-day remand period.
SO ORDERED.