This appeal is from the March 28, 1980, Food and Drug Administration ("FDA") final order, 45 Fed.Reg. 20,559, denying a request by Marshall Minerals, Inc. ("Marshall"), for a public hearing on its petition, dated August 12, 1977, for a food additive regulation pursuant to sections 409(b) and (c) of the Federal Food, Drug and Cosmetic Act, 21 U.S.C. §§ 348(b) and (c).1 We reverse and remand for a public hearing pursuant to 21 U.S.C. § 348(f)(1), which provides for a public hearing on objections to an order by any person adversely affected thereby and action thereon by the Secretary by order made public.BACKGROUND

Much of the relevant background for this case is reported in Southeastern Minerals, Inc. v. Harris, 622 F.2d 758 (5th Cir. 1980), and United States v. An Article of Food Consisting of 345/50-Pound Bags, 622 F.2d 768 (5th Cir. 1980), and need not be repeated here.

Marshall's petition relates to its gentian violet premix poultry feed. After the petition was filed and after the district court in Southeastern Minerals, supra, ruled that FDA had violated its own rules by failing promptly to notify Marshall of the disallowance of the petition, as noted by this court in Southeastern Minerals, supra at 762, FDA accepted the petition and published the proposed regulation.2 On May 8, 1978, however, FDA notified Marshall by letter that the stability data required under 21 C.F.R. § 571.1(A) still had not been received.3 In this letter FDA first stated that its denial of the regulation was based on deficiencies in four areas: safety, utility, stability, and Environmental Impact Analysis Report ("EIAR"). FDA's position with respect to these four areas not only framed the issues in the subsequent proceedings before FDA but is crucial to resolution of the issue in this appeal. The letter stated that the data contained in Marshall's petition had been reviewed, but were inadequate for the following reasons:

The submitted short-term toxicity studies reveal no clear adverse production effects of feeding gentian violet to chickens, however there were no histological examinations of tissues from the treated birds in those studies. The short-term duration of these studies and the lack of histological examination precludes assessment of the demonstrated carcinogenic potential of gentian violet .... The submitted residue data are identical to those data submitted by Naremco in 1974 (4 and conflict with other more recent data which indicate residues of gentian violet in edible tissues and eggs and which were derived from studies using significantly more sensitive detection methods.

1. The studies reported in this submission concerning the inhibitory effects of gentian violet on growth of Aspergilli and other fungi and the biosynthesis of aflatoxins are not adequate to support the claim of utility in poultry feed. Most of these data have been generated from studies in which laboratory media were used as the substrate for fungal growth. The utility of gentian violet must be demonstrated.

2. The Somporn Study, (5 part of which was conducted in feed, indicated that crystal violet had no inhibitory effect on Aspergilli in feed when the dye was added at 150 ppm. This concentration is approximately ten times the proposed use level.

3. The submitted report by Kingsland and Anderson does not qualify as supportive data for utility because of experimental deficiencies.

a. No zero time samples were taken to determine the level of contamination of the feed prior to inoculation with the test organisms.

b. Experimental conditions during the 12 week feed incubation period were not specified.

c. Isolates from the uninoculated controls were not identified. Therefore, the fungi growing in the feed could have been any fungi including those used to inoculate.

d. If the fungal counts from the uninoculated controls are compared after 12 weeks of incubation it is evident that a tremendous amount of variation occurred among them (range of 4.3 X 10 3 to 1.9 X 10 7). This variation in background contamination makes it impossible to draw conclusions about the effectiveness of gentian violet on the test organisms because the proportion of total counts due to contaminants in the treated feed is unknown.

As noted above and in our letter of January 23, 1978, (6 another inadequacy in your petition is the lack of stability data specified under and required by 21 C.F.R. 571.1(A).

(D. EIAR is covered in Conclusion 4, infra.)

1. Properly designed and carefully conducted and controlled chronic toxicity studies in test animals must be performed and completed. Studies reported in your submission are inadequate to determine long-term effects of feeding gentian violet. Therefore, it is impossible to evaluate the safety to humans from the use of your product under the conditions of intended use.

2. Adequate studies to demonstrate the utility of gentian violet as (a) fungistat in feed above have not been conducted.

3. It is apparent from studies reported elsewhere that fungi differ considerably in their sensitivity to gentian violet. Therefore, the fungi against which gentian violet is effective at the proposed concentration for use will have to be specified.

4. The (EIAR) submitted with your petition is incomplete. See Title 21 CFR 25.1. Specific data supporting safety of gentian violet and specific details of the blending process were not included. Final analysis of your EIAR is not possible until the pending questions on safety have been resolved and until the details of your blending process have been submitted.

Pursuant to the decision of the District Court, entered on April 19, 1978, your petition has been deemed "filed". Notice of the regulation you have proposed will be published in the Federal Register within 30 days of April 19, 1978. This letter constitutes a determination, on the merits of the data you have submitted, that your petition is inadequate and, as submitted, would not support such a Food Additive regulation. Consistent with agency policy and in accord with the Court's decision, you have thirty days to advise us whether, and if so, how you intend to satisfy the serious deficiencies in your petition. In the absence of the required data, an order consonant with the data submitted will be issued.

In response to FDA's letter, Marshall stated in a letter dated July 7, 1978, that it intended to maintain its position that gentian violet in general as well as in its premix is not subject to the Food, Drug and Cosmetic Act because gentian violet is generally recognized as safe. See United States v. An Article of Food Consisting of 345/50-Pound Bags, 622 F.2d at 770.

In regard to the alleged safety deficiencies Marshall declined to respond, requesting definition from FDA for the terms "short-term" and "long-term" in the context of gentian violet toxicity studies as well as copies of the "other more recent data" referred to in FDA's May 8, 1978, letter. Marshall indicated that it would respond after it had reviewed this requested information.

In regard to the alleged utility deficiencies Marshall stated that the "mere fact that some of the work has been generated from studies in which laboratory media were used as the substrate for fungal growth does not preclude their value as supported (sic supportive) data for such utility claim." Marshall also noted "that the sections of Food and Drug Administration law regarding the food additive petition are directed more toward safety than toward utility ... (and assumed) ... that any problems regarding utility would work themselves out in the market place, for certainly if the product does not work the consumer will not purchase it."

Regarding the "Somporn Study" comments in the May 8, 1978, FDA letter, Marshall explained "that the claim of Marshall Minerals is not for specificity of action against Aspergilli, but as an aid in the control of fungus and mold growth in poultry feed." Marshall pointed to data in the "Somporn Study" on four fungi or molds under varying conditions with gentian violet concentrations of from 15-80 parts per million ("ppm") showing growth inhibitions ranging from partial to complete, and concluded from these data that its utility claim was clearly supported.7

As to the report of Kingsland and Anderson, Marshall submitted a letter and statement by Mr. Graydon C. Kingsland, Associate Professor in the Department of Plant Pathology and Physiology at Clemson University and coauthor of the study, the relevant portions of which follow:

a. Zero time values for propagules of the test fungi in the feed are the values labeled "Unaltered Controls" in Table 2. These values were obtained from dilutions of the unaltered feed. Relatively little, if any, growth or sporulation would be expected at 8.5% moisture. These values, consequently, are probably good approximations of the level of original contamination.

b. The samples were stored under general laboratory conditions with temperatures between 70o and 76o F. This is near the optimum temperature for growth and development of F. moniliforme. It is 8o to 10o below the optimum for A. flavus, although A. flavus will develop rapidly at the storage temperature employed in this work.

c. All colonies of A. flavus were identified as such. All colonies isolated from unaltered controls were from naturally occurring A. flavus in the original feed, since no inoculum was added to these control samples. A. flavus colonies from the artificially contaminated treatments and controls were total colonies some natural background, and some inoculated to the feed, as described.

d. The variation in counts was not unusual and, in fact, could be expected for F. moniliforme under the storage conditions employed. Relatively little development and sporulation would be anticipated for F. moniliforme at 8.5% moisture. Consequently, the population (4.3 X 10³) in the feed probably approximates the original population. At 19% moisture, on the other hand, growth and sporulation occurred in the stored feed, resulting in the much higher populations (67.6 X 10 3).

In response to "the lack of stability data specified under and required by 21 C.F.R. 571.1(A)," Marshall submitted data from an analysis by Woodson Tenent Laboratories on samples of all gentian violet premix which had been produced by Marshall. In this submission were comparisons of the percentage of gentian violet found in each of 18 batches at the time each batch was made to the percentage of gentian violet found in retained samples of each batch as of April 3, 1978. The initial range of percentage of gentian violet was from 3.18% to 1.45% and the range on April 3, 1978, was from 1.78% to 1.51%. Along with the raw data, the accompanying memo explained that

(t)hroughout the entire test period, there tends to be a direct relationship between the original assay and the assay of April 3, 1978. In some cases, assaying slightly higher in 1978 than the original assay and in other cases, assaying slightly lower; but all within deviations of acceptable standard.

In regard to the EIAR, Marshall considered specific details of its blending process a trade secret but did set forth its general blending process:

The gentian violet is disbursed on calcium carbonate carrier to give a 1.6% gentian violet level in the final mixture by blending gentian violet with an inert polar, liquid solvent disbursing this liquid mixture in a horizonel (sic) batch mixer containing the calcium carbonate. This process is carried out in a mixer which includes the latest and most modern dust controlled equipment in the feed industry.

In a letter dated September 19, 1978, and considered to be a final determination on the merits of Marshall's petition, FDA again concluded that Marshall's petition remained inadequate, based on deficiencies in the same four areas. The reasons stated were substantially identical to those in its May 8, 1978, letter; however, the September 19, 1978, letter was more specific in some instances.

First, with regard to the safety data, FDA noted that " 'long' term is synonomous with 'chronic' and 'lifetime' testing ... (and that) ... (o)ur conclusion on page 2 of our May 8 letter specifically and clearly calls for the need to conduct carefully controlled toxicity tests regarding gentian violet before your product can be approved." The letter goes on to explain that, although Marshall's residue data were accurate to .2 ppm, other data8 showed "that gentian violet or its metabolites may be present in the edible parts of animal tissue down to 50 ppb. (parts per billion) and may be present at even lower levels." This possibility raised safety concerns of FDA because there were "no data constituting an adequate toxicity base upon which the safety of residues of gentian violet known to be present in edible animal tissues can be evaluated." FDA pointed out that Marshall submitted no information on the

total residue of gentian violet that may remain in the edible parts of animal tissue ... (and) ... the residue studies submitted were not run on your (Marshall's) product, but rather were run three years before you even began to manufacture a gentian violet product. The product that was tested was GV-11, a product manufactured by Naremco, Inc. Accordingly, even if this study were of the requisite sensitivity, data would have to have been submitted demonstrating the equivalency of your product and GV-11.

FDA stated that the utility data "fail for two reasons to support the claim that gentian violet inhibits the growth of fungi when used as proposed in animal feed." The first was that Marshall's product contained the crystalline form of the compound, whereas the gentian violet tested was in solution; the second was that Marshall's product was to be used in animal feed, although the test data were generated from use of laboratory nutrient systems.

FDA advised that Marshall's arguments regarding the Somporn study were unpersuasive; that Somporn demonstrated that crystal violet9 had no inhibitory effect on Aspergilli mold present in feed when crystal violet was added at 150 ppm, a level nearly ten times Marshall's recommended dosage.

In response to Marshall's statements concerning the report by Kingsland and Anderson, FDA remarked that the experiments in that study "were not properly controlled (because they) ... were not sterilized and in all probability contained many different types of fungi"; and that "the reduction in the number of viable organisms may have been due to (a) low moisture level and not to the presence of gentian violet...."

As to Marshall's statements concerning use of the feed for many commercially processed chickens without complaint, FDA said that "a compound may have absolutely no anti-fungal effects in feed and at the same time produce no overt negative effects on the birds."

FDA again maintained that Marshall's stability data failed to comply with 21 C.F.R. § 571.1, and, this time, stated its reasons, as follows:

First, no data were submitted to verify the chemical identity and composition of the gentian violet, the food additive, you employ. This deficiency could be rectified by a submission describing the components and composition of the gentian violet employed as well as a submission describing the synthesis of the gentian violet as performed by your supplier E.I. Dupont DeNemours and Co., Inc. Furthermore, a statement regarding the minimum content and identity of the components of reaction by-products and other impurities in your gentian violet food additive should have been submitted. Without these data it is impossible to show that the active gentian violet food additive is prepared according to a recognized standard, such as U.S.P. XIX gentian violet. Finally, a representative batch formula of your gentian violet food additive should also be submitted.

The second major area of deficiency in your stability data regards the failure to provide any information demonstrating that your product in its market container as well as in representative types of feed, including pellets, is stable, i. e., is insured to possess a uniform identity, strength, quality, and purity for a given period of time. Such data should be collected from appropriately designed experiments in which anticipated or reasonably expected environmental conditions of storage are used. These studies should include storage at room temperature as well as at 38o C. for at least 180 days for a premix and for not more than 90 days for the complete broiler feed. Furthermore, the data collected from these experiments should be analyzed to assess the need for an expiration date for the product. No discussion whatsoever of an expiration date was provided in your submission. Finally, in order to assure that stability is maintained during manufacturing, you must make some provision for sampling and analysis of the first five production lots of premix followed by five percent of the annual production thereafter made.

The third major area of deficiency regarding your submission regards the failure to submit specific information regarding not only the methods but also the facilities and personnel used and employed in each part of the manufacturing process and packaging operation of your product. Any such information would by necessity also contain a description of the analytical controls used at each of the various stages of manufacturing. Because your stability data were lacking in several other regards, we are enclosing a copy of our stability guidelines, recently prepared, which should provide what apparently is necessary clarification of the stability requirements under 21 CFR 571.1.

Finally, with respect to the EIAR, FDA stated that

no EIAR can be considered acceptable until all human safety questions concerning gentian violet have been satisfactorily answered. Furthermore, the EIAR must contain detailed information about the method used to control wastes resulting from manufacturing your gentian violet premix and how these controls meet current local, state and federal pollution control requirements.

The design, implementation, and analyses of the requisite chronic long-term toxicity tests required to demonstrate the safety of gentian violet for use in animal feed entail at least a period of three years. It would not be only against the interest of the public protected by section 409 but also unlawful under that section to allow, as suggested by Mr. Kirbo, a food additive petition regarding a suspect carcinogen to be unacted upon for such a substantial period. Accordingly, this letter constitutes a final determination, on the merits of the data you have submitted, that your petition is inadequate and as submitted does not support a food additive regulation. An order consonant with this determination will be issued and published in the Federal Register. The opportunity to contest our decision that "chronic"/" long" term testing is required as well as the opportunity to contest any of our decisions regarding the adequacy of the data you have submitted is available to you in the form of an administrative scientific hearing pursuant to 21 U.S.C. 348(f)(1).

The promised order was published on March 30, 1979, 44 Fed.Reg. 19,035, and contained a brief procedural history of the petition as well as a discussion of numerous studies bearing on the effect of gentian violet and related compounds on human and various animals. Included in the order were reasons for denial of Marshall's feed additive petition, which reasons were substantially identical to those in FDA's September 19, 1978, letter to Marshall.

In response, Marshall filed 24 objections and requested a section 348(f)(1) hearing on each and additional time to amend and supplement its objections. FDA granted the request for additional time because of the "amount of scientific material which must be reviewed and evaluated." 44 Fed.Reg. 40,93 3 (1979). However, no amendment or supplement was filed by Marshall. On March 28, 1980, FDA denied Marshall's request for a hearing, 45 Fed.Reg. 20,559, and this appeal followed.

The issue is whether FDA erred in summarily disposing of Marshall's food additive petition without a public hearing. Public hearings on food additive petitions are governed by section 348(f)(1) of the Federal Food, Drug, and Cosmetic Act, as amended (21 U.S.C.A. § 348(f)(1)), which provides:

(f)(1) Within thirty days after publication of an order made pursuant to subsection (c) or (d) of this section, any person adversely affected by such an order may file objections thereto with the Secretary, specifying with particularity the provisions of the order deemed objectionable, stating reasonable grounds therefor, and requesting a public hearing upon such objections. The Secretary shall, after due notice, as promptly as possible hold such public hearing for the purpose of receiving evidence relevant and material to the issues raised by such objections. As soon as practicable after completion of the hearing, the Secretary shall by order act upon such objections and make such order public.

Although section 348(f)(1) provides for a public hearing for persons adversely affected by a section 348 order, FDA may summarily deny such a hearing, at least insofar as the order relates to safety issues, " 'where it is apparent at the threshold that the applicant has not tendered any evidence which on its face meets the statutory standards as particularized by the regulations.' " American Cyanamid Co. v. Food & Drug Administration, 606 F.2d 1307, 1312 (D.C.Cir.1979) (quoting from Weinberger v. Hynson, Westcott & Dunning, Inc., 412 U.S. 609, 620, 93 S. Ct. 2469, 2478, 37 L. Ed. 2d 207 (1973).

Further, when FDA uses a case-by-case particularization of statutory safety standards, as it has on gentian violet as a food additive,

... it is clearly incumbent upon it to give the applicant notice of those standards and of the manner in which the data before it failed to meet them. This notice must be given in timely fashion to put the applicant in position to dispute FDA's interpretation of the Act's safety criteria, object to FDA's critique of the submitted studies, and conduct and proffer new studies meeting the newly-articulated requirements. Should the applicant then identify a material issue of fact, FDA must hold a hearing. (Footnote omitted and emphasis added.)

American Cyanimid Co. v. Food & Drug Administration, 606 F.2d at 1314. Evidence on the safety-related aspects of Marshall's petition consists of studies submitted by Marshall and studies cited by FDA. From these studies, it must be determined whether there is a material issue of fact concerning the safety of Marshall's gentian violet feed additive.

Section 348(c)(3)(A), supra, provides that no FDA regulation shall issue if a fair evaluation of the data before it fails to establish that its use will be safe. There is difficulty in ascertaining the meaning attached to the term "safe" by FDA in its administration of the statute at the time it was reviewing Marshall's petition.10 In defining "safe" or "safety" relative to food additives, 21 C.F.R. § 570.3(i) currently provides:

(i) "Safe" or "safety" means that there is a reasonable certainty in the minds of competent scientists that the substance is not harmful under the intended conditions of use. It is impossible in the present state of scientific knowledge to establish with complete certainty the absolute harmlessness of the use of any substance. Safety may be determined by scientific procedures or by general recognition of safety. In determining safety, the following factors shall be considered:

(1) The probable consumption of the substance and of any substance formed in or on food because of its use;

(2) The cumulative effect of the substance in the diet, taking into account any chemically or pharmacologically related substance or substances in such diet;

(3) Safety factors which, in the opinion of experts qualified by scientific training and experience to evaluate the safety of food and food ingredients, are generally recognized as appropriate.

Further, 21 C.F.R. § 570.20 sets forth the following general principles for evaluating the safety of food additives:

(a) In reaching a decision on any petition filed under section 409 of the act, the Commissioner will give full consideration to the specific biological properties of the compound and the adequacy of the methods employed to demonstrate safety for the proposed use, and the Commissioner will be guided by the principles and procedures for establishing the safety of food additives stated in current publications of the National Academy of Sciences-National Research Council. A petition will not be denied, however, by reason of the petitioner's having followed procedures other than those outlined in the publications of the National Academy of Sciences-National Research Council if, from available evidence, the Commissioner finds that the procedures used give results as reliable as, or more reliable than, those reasonably to be expected from the use of the outlined procedures. In reaching a decision, the Commissioner will give due weight to the anticipated levels and patterns of consumption of the additive specified or reasonably inferable. For the purposes of this section, the principles for evaluating safety of additives set forth in the above-mentioned publications will apply to any substance that may properly be classified as a food additive as defined in section 201(s) of the act.

(b) Upon written request describing the proposed use of an additive and the proposed experiments to determine its safety, the Commissioner will advise a person who wishes to establish the safety of a food additive whether he believes the experiments planned will yield data adequate for an evaluation of the safety of the additive.

Marshall points out that in August 1977, when its petition was filed, FDA regulations defined safe to be whether there was no "substantial risk of harm," but that in October 1977 the definition was reworded to establish a higher standard for safety, as set forth in 21 C.F.R. § 570.3(i), supra. See also 41 Fed.Reg. 38,644 (1976), as amended at 42 Fed.Reg. 55,206 (1977). FDA asserts that this later definition is taken from the legislative history of the Act. See H.R.Rep.No.2284, 85th Cong., 2d Sess. 4-5 (1958). We are satisfied that, even under the current standard of safety as defined in 21 C.F.R. § 570.3(i), there are material issues of fact, for which a section 348(f)(1) hearing must be held.

(A) Carcinogenicity Concerns

Regarding the safety of Marshall's chicken feed additive, FDA's reasons for denial of the petition relate especially to the potential carcinogenic effects of gentian violet on the chickens themselves and on humans from gentian violet residue in edible parts of the chickens. However, it appears that FDA has ignored the statutory requirements for considering a feed additive petition when cancer is suspected. Thus, as quoted earlier, while section 348(c)(3) provides that no additive should be deemed safe if found to induce cancer when ingested by man or animal, it further provides that an additive shall not be deemed unsafe in the case of feed for animals raised for food if the Secretary finds that the additive will not adversely affect the animals and that no residue of the additive will be found in any edible portion of such animals. There is no finding by FDA that, under the conditions of use specified for Marshall's chicken feed additive, gentian violet induces cancer when ingested by man or animal; nor is there any finding of whether or not, under such conditions, gentian violet adversely affects chickens or will be found in any edible portion of chickens.

On the question of potential carcinogenic effects of the additive on chickens themselves, Marshall relies on several studies,11 which show the effects of feeding various concentrations of gentian violet to chickens. From these studies, and one in particular which stated that there was no significant risk of harm to chickens fed gentian violet at concentrations of up to 20.625 ppm, Marshall concluded that use of its food additive is safe because it is fed to chickens at a lower concentration, viz., 8 ppm. Here, FDA's objection is twofold: (a) that the studies were "short-term" and (b) that none contained histological examinations of tissues from the treated chickens to counter recent data, derived with the aid of more sensitive detection methods, which FDA said indicated residues of gentian violet in edible tissue and eggs. The histological examinations are, in FDA's view, required because of "the demonstrated carcinogenic potential of gentian violet." FDA amplified its statement of "carcinogenic potential," thus:

Although there are no published studies to date that demonstrate that gentian violet causes cancer when ingested, unpublished studies conducted by Fitzhugh in 1949 indicate that long-term feeding of gentian violet to rats is followed by the appearance of neoplastic nodules and hyperplastic foci in the liver, sometime coexistent with hepatocellular carcinoma suggesting that the carcinoma may have been induced by gentian violet ....

Additional concerns regarding the carcinogenicity of gentian violet arise from its chemical structure. Crystal violet, a main component of gentian violet, belongs to a class of dyes related by molecular structure, which can be generally referred to as aminophenylmethanes. A few of these are known animal carcinogens, and two, auramine and magenta, have been implicated as human carcinogens as a result of observations of human exposures in the dye industry ....

Further information regarding the potential carcinogenicity of gentian violet is found in a lecture given at Yale University in 1939 on studies with 50 compounds. Kinosita claimed that hexamethyltriaminotriphenylmethane, a reduced form of crystal violet, caused gastric papillomas and slight adenomatous proliferation in the liver when given orally to rats over a period of 300 days .... No dose or other experimental details were given.

44 Fed.Reg. 19,036 (1979).

The Fitzhugh "studies"12 involved the effect on various tissues of 52 male and 48 female rats from ingestion of various concentrations of gentian violet. Gentian violet was fed to the rats as follows:

The Rodricks report concluded from the Fitzhugh data that "(n)o treatment related lesions were observed in any organ or tissue except the liver" and that none of the three types of lesions13 "was observed in either male or female control rat livers." The report further concluded:

... One high dose female animal had coexistent neoplastic nodule and hepatocellular carcinoma which suggests that the carcinoma may have been induced by gentian violet. The lesions described are not cancerous or clearly pre-cancerous but represent neoplastic-type alterations which are interpreted as qualitatively in the direction of cancer.

Despite a number of significant deficiencies in the design and execution of the study, the positive observations contained in this report are not obviated by the deficiencies.

Therefore, it cannot be concluded that the chronic feeding of gentian violet to rats is innocuous since dose-related neoplastic-type proliferations were observed in their livers. The precise relationship of the lesions with respect to carcinogenesis will require further study to elucidate.

Although it is clear from these studies that relatively high dose rates of gentian violet will cause lesions "in the direction of cancer" in the livers of rats, particularly females, we are not persuaded that they support FDA's summary disposition of Marshall's petition because of a "demonstrated carcinogenic potential of gentian violet" with respect to the much lower dosages of Marshall's chicken feed additive. American Cyanamid Co. v. Food & Drug Administration, 606 F.2d at 1312-14.

In support of its "additional concerns regarding the carcinogenicity of gentian violet," FDA cited three studies14 which implicate, as human carcinogens, two compounds related to gentian violet by their chemical structure. These two compounds, auramine and magenta, as well as gentian violet, are aminophenylmethanes, a class of compounds having widespread use as dyes.

The Williams study, note 14, supra, reviewed the effects on body tissues of rats and mice fed various dosages of auramine and concluded that auramine "is carcinogenic to the liver of rats and mice when administered orally in large doses." However, there are no data in the Williams study whereby the phrase "large doses" may be objectively determined. The Case study evaluated the risk of bladder cancer to workmen engaged in the production of auramine or magenta in the British chemical industry between 1910 and 1952. It noted that a number of process steps occurred and a number of intermediate chemicals were produced in the overall process of making auramine and magenta. It concluded that the process had "a definite occupational hazard of causing tumour of the urinary bladder attached to it ..." but that the "analysis is not capable of determining the causal factor in this risk attached to the manufacture of magenta and auramine, and it should not be assumed that the finished products are necessarily dangerous."

The Arcos study discusses the carcinogenic potential of twelve di- and triphenylmethane amines. Of the twelve compounds studied, two are closely related to auramine in structure. One of these two is dimethylaniline, the starting material in the manufacture of auramine, and the other is 4,4'-tetramethyldiaminobenzophenone, an intermediate compound produced in the manufacture of auramine. This study noted the history of bladder cancer as an occupational hazard among workers employed in the manufacture of auramine and then stated that no tumors were found in rats after repeated injection of dimethylaniline. Arcos found, however, that "papillomas of the stomach and neoplastic changes in the liver" resulted from oral administration of 4,4'-tetramethyldiaminobenzophenone. The remainder of the study contains a discussion of the other ten amines and attempts to develop a theory whereby the carcinogenicity of species of this class of compounds may be predicted based on certain structural characteristics of the species.

Accepting these reports at face value, it is clear that gentian violet causes certain types of lesions in rats when administered in high dosages. What is not clear is (a) whether the types of lesions induced are intended to be included as section 348(c)(3)(A) cancers; (b) whether the effects of ingestion of gentian violet on chickens or humans would be the same as the effect on rats; (c) whether the effects of ingestion of gentian violet at high dosages may be extrapolated to corresponding effects at low dosages, such as those resulting from the conditions of use specified for Marshall's premix; and (d) whether the effects from ingestion of compounds other than gentian violet, notwithstanding their structural similarity to gentian violet, may be validly attributed to gentian violet.

Bearing in mind the statutory requirement, set forth in section 348(c)(3)(A), that FDA is to make an evaluation of whether the food additive, under the conditions of use, induces cancer when ingested by man or animal, and the Cross study,15 which directly supports Marshall's claim that gentian violet is safe when ingested under its recommended conditions of use, we are satisfied that there are material issues of fact regarding the carcinogenicity of gentian violet at the dose rates recommended for Marshall's product.16

In FDA's final order, alleged deficiencies in the length of time and degree of sensitivity of the carcinogenicity studies formed separate bases for denial of Marshall's petition. If carcinogenicity remains an issue at a public hearing, it may be necessary to further develop these points.

(B) DNA Damage Concerns

FDA stated in the "safety" section of its order that "(g)entian violet is also known to produce various chromosomal anomalies in cultured mammalian cells," that it is "capable of inducing damage to DNA ... (and that) the ability to produce chromosome anomalies and induce DNA repair, are two phenomena inducible by gentian violet treatment for which high positive correlations with carcinogenicity have also been claimed." FDA acknowledged, however, that "gentian violet does not produce apparent genetic damage to chick embryos"; and, despite the fact that "(t)he most successful correlations with carcinogenic activity have been carried out using data from point mutation assays ... (g)entian violet ... appears to be negative in these assays."17 Moreover, the Cross study, supra, states that gentian violet, fed at higher dose rates than recommended for Marshall's premix, is safe for chickens. Accordingly, we believe there is a material issue of fact over the safety of gentian violet in Marshall's premix with respect to its potential to damage DNA.

FDA also denied Marshall's petition on the ground that "the effectiveness of gentian violet in inhibiting the growth of mold in animal feed has not been established." Although numerous references and arguments have been advanced by Marshall and FDA in regard to the effectiveness of gentian violet on a chicken feed additive, it is unnecessary to reach this issue because there is no statutory basis for denial of a food additive petition on the ground of lack of effectiveness per se. FDA approval for both new drugs and food additives requires proof that the drugs or additives are "safe." New drugs must, additionally, be shown to be "effective." Weinberger v. Hynson, Westcott & Dunning, Inc., 412 U.S. 609, 93 S. Ct. 2469, 37 L. Ed. 2d 207 (1973), American Cyanamid Co. v. Food & Drug Administration, 606 F.2d 1307 (D.C.Cir.1979), United States v. Articles of Food and Drug Consisting of Coli-trol 80, 518 F.2d 743 (5th Cir. 1975). Since FDA's objection on this ground has no foundation in any specific statute or regulation, a summary disposition without a public hearing was improper.18 American Cyanamid Co. v. Food & Drug Administration, 606 F.2d at 1313.

Although section 348 does not make a food additive's stability a basis for denial of a food additive petition, 21 C.R.F. § 571.1(a) requires

A. The name and all pertinent information concerning the food additive, including chemical identity and composition of the food additive, its physical, chemical, and biological properties, and specifications prescribing the minimum content of the desired component(s) and identifying and limiting the reaction byproducts and other impurities. Where such information is not available, a statement as to the reasons why it is not should be submitted.

....

The petition shall include stability data, and, if the data indicate that it is needed to ensure the identity, strength, quality, or purity of the additive, the expiration date that will be employed.

FDA alleged in its order of March 30, 1979, 44 Fed.Reg. 19,035, that Marshall failed

to submit data to verify the chemical identity and composition of the gentian violet ... (and) to provide any information demonstrating that the proposed product in its market container as well as in representative types of feed, including pellets, is stable, i. e., will possess a uniform identity, strength, quality, and purity for a given period of time. Such data should be collected from appropriately designed experiments in which anticipated or reasonably expected environmental conditions of storage are used.

Marshall relies on its original petition "to verify the identity of its feed additive." Therein were included the chemical structure of gentian violet and copies of several patents relating to gentian violet used in feed additives, together with information from its supplier such as specifications and test data. With respect to stability of its food additive, Marshall, in a letter dated July 7, 1978, submitted to FDA "stability test data which has been run by Marshall ... on retained samples of all gentian violet pre-mix which it has ever produced." The analysis covered a three-year period and compared the percentage of gentian violet in the original batch with the percentage as of April 3, 1978. This would appear to fully comply with 21 C.F.R. § 571.1 in the absence of other bases of objection by FDA.19 Therefore, FDA's denial of Marshall's petition on the ground that it failed to comply with 21 C.F.R. § 571.1 was also improper.

Finally, FDA denied Marshall's petition on the ground that its EIAR20 was unacceptable. According to FDA "(n)o EIAR can be considered acceptable until all human safety questions concerning gentian violet have been satisfactorily answered." FDA also objected to the EIAR because it did not contain "detailed information about the method used to control wastes resulting from manufacturing the proposed gentian violet premix and how these controls meet current local, State, and Federal pollution control requirements ...."

With respect to its contention on "human safety questions," FDA has stated no reason related to environmental considerations, per se, for its denial of Marshall's petition.21 Although it is possible that FDA might establish separate grounds for denial of a food additive petition based on safety to the environment, it has failed to set forth any such separate grounds on this record. Accordingly, the safety concerns alluded to by FDA in this section of its final order are presumed to be identical to those raised by FDA under section 348(c). As pointed out, supra, there are material issues of fact related to those safety concerns.

With respect to the "detailed information" requested by FDA, Marshall, in its original petition, filed an EIAR and explained that its process was one of blending, not manufacture; that its blending process used state of the art equipment which resulted in no wastes; and that, consequently, there were no local, state, or federal pollution control requirements to be met. Accepting these submissions at face value, it is clear that not only was 21 C.F.R. § 25.1 complied with, but also that no separate environmental safety concerns prevented approval of Marshall's petition.

In its final order of March 28, 1980, and in its brief before us, FDA argues that a section 348(f) hearing should be denied Marshall because, in response to FDA's March 30, 1979, order dealing with the merits of the petition, Marshall's 24 objections were merely "legal argument and legal conclusions in the place of a reasoned and articulated response" to the safety issues raised in said order. According to FDA, 21 C.F.R. § 571.10 and 21 C.F.R. §§ 12.22 and 12.24 require

any petitioner for whom a food additive petition has been denied and who desires a hearing to submit reasons why the application should not be denied together with well-organized full factual analyses of the data the petitioner is prepared to provide in support of its opposition to the notice of denial. When it clearly appears from the data and from the reasons and factual analyses in the request for hearing that there is no genuine issue or substantial issue of fact, FDA may deny a hearing. What the agency in effect has said is that it will not provide a formal hearing where it is apparent at the threshold that the applicant has not tendered any evidence which on its face meets the statutory standards as particularized by the regulations. The propriety of this procedure has been expressly upheld by the Supreme Court. See Weinberger, et al. v. Hynson, Westcott, and Dunning, 412 U.S. 609, 620-621 (93 S. Ct. 2468, 2478-79, 37 L. Ed. 2d 207) (1973); see also United States v. Storer Broadcasting Co., 351 U.S. 192, 205 (76 S. Ct. 763, 771, 100 L. Ed. 1081) (1956).

However, as pointed out earlier, when a petitioner identifies either a material issue of fact, or, a FDA ground for denial having no foundation in any specific statute or regulation, he is entitled to a public hearing. In this case, not only were material issues of fact identified prior to issuance of FDA's March 30, 1979, order, but at least some of Marshall's objections to that order appear to be well taken.22

In view of all the foregoing, FDA's final order of March 28, 1980, is VACATED, and Marshall's food additive petition is REMANDED for a public hearing in accordance with section 348(f)(1).