MR. JUSTICE BRENNAN delivered the opinion of the Court.

Commercially grown Florida and Texas oranges have for many years been colored with a red coal-tar color. In 1939 the Food and Drug Administration, after testing and pursuant to § 406(b) of the Federal Food, Drug, and Cosmetic Act, certified this color, FDC Red No. 32 (hereafter Red 32), to be harmless and suitable for use in food. However, the Secretary of Health, Education, and Welfare, on November 10, 1955, ordered Red 32 and two other coal-tar colors to be removed from the certified list, after new tests in 1951-1953 cast doubt whether Red 32 was harmless, and after public hearings were held upon the matter on notice published in the Federal Register. The consequence of the Secretary's order was that under § 402(c) of the Act any food bearing or containing such colors would be deemed to be adulterated.

The validity of the Secretary's order was attacked in petitions under § 701(f) of the Act filed in several Courts of Appeals by persons and organizations claiming to be adversely affected. The Court of Appeals for the Second Circuit sustained the order against a general attack. Certified Color Industry Comm. v. Secretary of Health, Education and Welfare, 236 F.2d 866. In the instant case, however, the Court of Appeals for the Fifth Circuit, by a divided vote, set aside the order insofar as it removed the certification of Red 32 as harmless and suitable for use as external coloring on Florida and Texas oranges. 246 F.2d 850.

The Secretary did not determine that Red 32 in the quantities used in color-added oranges was harmful for human consumption, but rather determined on the basis of the 1951-1953 tests only that Red 32 and the other suspect coal-tar colors were toxic and therefore not "harmless and suitable for use in food." The Court of Appeals held that the 1939 finding that Red 32 was harmless "should not be supplanted" by a contrary finding "unless there is evidence that, in the amounts used, and in the manner of use, oranges colored with Red 32 are unsafe for human consumption." 246 F.2d, at 861-862. The word "harmless" was construed to be a term "of relation," preventing the Secretary from denying the continued use of Red 32 in the quantities used in color-added oranges in the absence of evidence that such quantities could not be consumed "without risk of injury or harm." Id., at 858. The Court of Appeals held further that in light of its premise that "harmless" was a term of relation and because two congressional Committees had found that the practice of adding the color to oranges was an economic necessity, it would be incumbent upon the Secretary to determine whether the use of the color was "required in the production" of food within the meaning of § 406(a), and if so, to promulgate a safe tolerance for Red 32 on oranges pursuant to that section. Until such a tolerance was promulgated, the court held that the Secretary was required to certify Red 32 as a safe color for use on oranges without one. Id., at 860-862. We granted certiorari to determine this controversial question of construction of this important statute designed for the protection of the public health. 356 U.S. 911.

Senate and House Committees have reported that the practice of adding color is an economic necessity in the production of Florida and Texas oranges for market. When mature oranges are removed from the tree, their skins, for botanical reasons unnecessary to detail here, are frequently green in color. Since the consumer would be prone incorrectly to interpret this greenness as a sign of immaturity, oranges are put through a "degreening" process which involves exposure to ethylene gas. In the case of certain California oranges, this gas process is sufficient to turn a green orange into one of the desired orange color. But the degreening process does not produce the desired color in Florida and Texas oranges; a light yellow shade results. The more desired color is therefore produced by immersing the oranges in, or spraying them with, a solution containing Red 32. The evidence at the hearings held by the Secretary was that the process infuses the peel of an orange with 0.0017% to 0.0034% of Red 32. Other evidence indicated that oranges taken as a whole, and candied peel, marmalade and orange juice would contain less — in many cases, much less — of the coal-tar color. It is conceded by the Secretary that there is no evidence that the level of ingestion of Red 32 involved in human consumption of color-added oranges is harmful.

However, the evidence at the Secretary's hearing did indicate that Red 32 had a poisonous effect on animals. Feeding the color to rats in quantities as small as 0.1% of their diet was deleterious and often fatal, with liver damage and enlargement of the heart in evidence. In larger quantities, 1.0% and 2.0% of the diet, ingestion of Red 32 by rats caused death within twelve days and a week, respectively. The health of dogs taking 0.2% of the color in their diets deteriorated rapidly; that of those taking 0.04% somewhat more slowly, but definitely; and ill effects were indicated at a feeding level as low as 0.01% of the diet. No safe level of administration of Red 32 to the test animals was established. These and similar tests, involving the administration of Red 32 and the other coal-tar colors involved to test animals generally as an item of diet, were the basis on which the Secretary's order rested.

The Secretary argues that the legislative history and the consistent administrative interpretation of the Act establish that his authority to list or continue the listing of coal-tar colors is confined to his authority under § 406(b) to certify "harmless" coal-tar colors, those which are wholly innocuous and demonstrated to be without adverse physiological effect. The argument runs that a toxic coal-tar color, such as the Court of Appeals agreed that Red 32 was, was to be prohibited completely without regard to whether it might possibly be used in safe amounts on a particular food product. The Secretary argues further that since Congress made known its will specifically and precisely in § 406(b) that a toxic coal-tar color, that is, one not "harmless," was not be certified under any circumstances, the tolerance provisions of § 406(a) have no relevance to the validity of his order.

We are of the opinion that the Court of Appeals erred and that its judgment cannot stand.

First. The provisions of §§ 402(c) and 406(b) dealing expressly with coal-tar colors were innovations in the Federal Food, Drug, and Cosmetic Act of 1938; there were no counterpart provisions in the original 1906 food and drug legislation. By these provisions, Congress carefully distinguished the treatment to be given by the Secretary to toxic coal-tar colors. The original Act dealt generally with poisonous and other deleterious substances in food, as are now treated under § 402(a), but it did not deal specifically with coal-tar colors. Section 7 of the original Food and Drugs Act, 34 Stat. 769, provided that an article of food should be deemed adulterated "If it contain any added poisonous or other added deleterious ingredient which may render such article injurious to health . . . ." This Court held in United States v. Lexington Mill Elevator Co., 232 U.S. 399, following the "plain meaning" of the statutory language, that this placed the burden upon the Government of establishing that the added substance was such as might render the food to which it was added injurious to health. This rule applied without distinction where coal-tar colors were involved. Congress was aware of the difficulties of this test which required that the questioned food product be evaluated as a whole, and of the existence in this area of an informal certification practice under the 1906 Act under which not food products but the coal-tar colors themselves were subjected to test to determine their poisonous or harmful character. Cf. S. Rep. No. 361, 74th Cong., 1st Sess., pp. 7-8. Of course, when litigation occurred, the Lexington Mill standard was applied. See W. B. Wood Manufacturing Co. v. United States, 286 F. 84, 86-87.

It was against this background that the 1938 statute was proposed and enacted. It is obvious to us that an approach different from the rule in Lexington Mill was intended by Congress when in § 402(c) and § 406(b) it addressed itself to the severable and narrow problem of coal-tar colors. The language involved in Lexington Mill survived generally in the Act's broadest and most general test of food adulteration, § 402(a)(1). Section 402(c) provided a separate test: that a food should be deemed adulterated "If it bears or contains a coal-tar color other than one from a batch that has been certified in accordance with regulations as provided by section 406 . . . ." Plainly Congress banned any addition to foods of coal-tar colors not certified by the Secretary. The standard established for the Secretary was set forth in § 406(b): "The Secretary shall promulgate regulations providing for the listing of coal-tar colors which are harmless and suitable for use in food and for the certification of batches of such colors . . . ." There appears in Senator Copeland's memorandum on the first of the bills which led to the 1938 Act, S. 1944, 73d Cong., 1st Sess., which contained new provisions on coal-tar colors similar to the ones in the Act as finally passed, a clear indication that one of the purposes of these provisions was to do away, in this area, with the Lexington Mill approach. 77 Cong. Rec. 5721. This had the effect of making the certification system, in which analysis concentrated on the color substances themselves, rather than an examination of the effect of the use of the colors in the context of the food products involved, the conclusive test of adulteration. Thus it is that the test of certification laid down in § 406(b) concentrates on the color substance itself; it is to be listed only if it is harmless. The Secretary is to address himself to the harmless character of the substance first; once this is assured, the statutory plan is that it may be freely used in foods, subject to the provisions of the other sections of the Act. Clearly such a plan is a rational one to ascribe to Congress. It is true that the ultimate purpose here concerned of the adulteration provisions of the Act is to protect health, and that no one makes the color substances by themselves an item of diet. But it certainly was competent for Congress, in the light of what were recognized problems to health in the use of such added colors, to adopt a rule of caution in treating this recognized and definable problem area. This rule of caution is here one which relieves the Secretary from the burden of showing in each case that a food containing them raises a possibility of injury to health, and requires that the color stuffs, whose positive values are only visual and which are not naturally found in foods, not be added unless they could pass a higher standard.

The significance of such an approach is demonstrated here. No safe level for ingestion of Red 32 has been established, either in respect of humans or of animals. No one contends that it is impossible that ill effects will be experienced in human beings if unrestricted use of the substance is permitted in articles of food. On the other hand, no instance of a harmful use of Red 32 in a particular food was established in the record. These questions present broad inquiries, difficult of proof, and doubtless apt to be more long-drawn-out in investigation than even the ones which the Secretary pursued. Yet it has been shown that the color of itself has poisonous properties. In the light of the over-all purpose of the Act, cf. United States v. Dotterweich, 320 U.S. 277, 280, and the specific terms here involved, it seems to us that Congress did not intend that a verdict of "not proven" on the questions mentioned should preclude the Government from preventing the use of substances like the one in question when they were shown to have poisonous effects by themselves.

We are not persuaded by the respondents' argument, adopted by the Court of Appeals, that the words "harmless" and "poisonous" are relative words, referring not to the effect of a substance in vacuo, but to its effect, taken in a particular way and in particular quantities, on an organic system. Of course this is so, but the question before us certainly does not depend on it. This is not a case like the examples put which remind us that pure water would be deleterious if taken at the rate of four gallons an hour or common table salt at several ounces. The color substances appear to have been administered at toxicologically significant levels; they played a relatively small part in the diets of the test animals, generally less, and frequently much less, than 1%. Obviously if the color substances themselves are made an item of diet in the trifling percentages used on the test animals, their effect is poisonous. Congress may have intended "harmless" in a relative sense, but we think it was in relation to such laboratory tests as the ones the Secretary performed that Congress was speaking when it required that coal-tar colors be "harmless." We do not believe that Congress required the Secretary first to attempt to analyze the uses being made of the colors in the market place, and then feed them experimentally only in the proportions in which they appeared in certain of the food products in which the colors were used. This appears to be the very procedure on which Congress turned its back in the 1938 Act.

The respondents contend that since the Secretary himself maintains various lists of certified colors, one containing colors harmless and suitable for all food, drug and cosmetic uses, another of colors harmless and suitable for general use in drugs and cosmetics, and a third of colors harmless and suitable for external use in drugs and cosmetics, 21 C.F.R. § 9.3, 9.4, 9.5, he has recognized that "harmless," as used in the statute, does not bear the "absolute" meaning he is alleged to give it. From this it is said to follow that the Secretary must, in forbidding the use of colors in foods, restrict his prohibition to specific food uses in which the color is shown to be capable of a deleterious effect. We do not draw this inference. Provisions similar to those dealing with the use of coal-tar colors in foods are repeated in the portions of the Act dealing with drugs and with cosmetics. Section 501(a)(4), 52 Stat. 1049, 21 U.S.C. § 351 (a)(4), proscribes the use of uncertified colors in drugs for the purpose of coloring, and refers to § 504, 52 Stat. 1052, 21 U.S.C. § 354, which authorizes the listing and certification of coal-tar colors which are "harmless and suitable for use in drugs." Comparable provisions are found with respect to cosmetics in §§ 601(e) and 604, 52 Stat. 1054, 1055, 21 U.S.C. § 361(e), 364. It is clear from these provisions that Congress contemplated that a color might be harmless in respect of drugs or cosmetics but not of foods. And the fact that the Secretary has established a further category, distinguishing between colors intended for external and for general use, we do not think inconsistent with our interpretation of the Act. These distinctions can be established through tests run on the color substance as such, in the way in which the Secretary has conducted the tests in the matter before us. It is a far cry from saying that the Act permits a generic distinction capable of laboratory proof, between external and internal uses of a color, to say that it commands that the colors cannot be inquired of at all except in the specific contexts of their use in food, drug and cosmetic products.

Second. But even if the Secretary's approach of viewing the harmlessness of coal-tar colors in terms of the colors themselves, rather than in their specific applications, is correct, the respondents insist, as the court below indicated, that the Secretary should establish tolerances for the use of colors in food, even though not found to be "harmless." The respondents point to § 406(a) of the Act which allows the Secretary to establish tolerances for poisonous substances added to food where the substance is "required in the production" of the food or "cannot be avoided by good manufacturing practice." They argue that this provision should be used by the Secretary to establish a maximum tolerance for the application of Red 32 to the skins of oranges, either because it applies by its own terms or it is applicable by analogy. The Secretary contends that he is without power to permit the use of harmful coal-tar colors in specific foods through a system of tolerances. We believe he is correct.

The Federal Food, Drug, and Cosmetic Act is a detailed and thorough piece of legislation. Its treatment of many public health and food problems is quite specific, and of course it is the duty of the courts in construing it to be mindful of its approach in terms of draftsmanship. Here again, in our construction of this explicit Act, we must be sensitive to what Congress has written, and recall that "It is for us to ascertain — neither to add nor to subtract, neither to delete nor to distort." 62 Cases of Jam v. United States, 340 U.S. 593, 596. Section 406(a), which provides for the system of tolerances, constitutes by its terms a definition of the term "unsafe," which appears in § 402(a)(2), a prohibition on foods which bear or contain "any added poisonous or added deleterious substance . . . which is unsafe within the meaning of section 406." This is a prohibition entirely separate and distinct from the prohibitions of § 402(c) on foods containing or bearing uncertified coal-tar colors. The existence of a tolerance is specifically stated in § 406(a) only to give sanction to what would otherwise amount to adulteration within the terms of § 402(a)(1). Accordingly, it is obvious from the language of the statute that the provisions authorizing the establishment of tolerances apply only to § 402(a)(1) and (2) and do not apply to § 402(c)'s flat prohibition against the use of uncertified colors. Respondents do not direct us to any substantial contrary indication in the legislative history. Nor can the tolerance provisions be applied to coal-tar colors through some form of analogy. The command of the statute is plain: where a coal-tar color is not harmless, it is not to be certified; if it is not certified, it is not to be used at all. In this regard also, an approach in terms of the toxicity of the coloring ingredient, rather than of the food product as a whole was chosen by Congress. It evidently took the view that unless coal-tar colors were harmless, the considerations of the benefits of visual appeal that might be urged in favor of their use should not prevail, in the light of the considerations of the public health. In the case of other sorts of added poisons, though only where they were required in the production of the food concerned or could not under good manufacturing practice be avoided, a different congressional policy was expressed in the 1938 enactment. It is the duty of the Secretary to give effect to this distinction; he has done so with apparent substantial uniformity and has done so here.

Third. After the promulgation of the Secretary's order, Congress afforded temporary relief to those economically interested in the coloring of oranges with Red 32. Legislation was enacted in the summer of 1956 to afford a period of approximately three years (until March 1, 1959) during which use of the color would be allowed solely in application to the skins of oranges. The statute does not, in our view, affect the situation presented to the courts for judicial review; the Secretary's order remains to be tested under the permanent provisions of the Act, insofar as they will affect respondents after March 1, 1959. The statute accordingly operates as a legislatively ordained stay of the Secretary's order insofar as it affects the present respondents and those similarly situated. See H.R. Rep. No. 1982, p. 3, and S. Rep. No. 2391, p. 3, 84th Cong., 2d Sess. In view of the very temporary nature of this legislative "stay," the automatic resumption of the status quo upon its expiration, and the effect of the order on the respondents, even during the legislative stay, we agree with the parties that the matter before us is not moot. The Secretary's order was the promulgation of a general rule, prospective in operation, and the facts of the respondents' business are such that if the order is upheld, there will be a practical effect on them even during the span of the temporary legislation. Accordingly, the respondents remain persons adversely affected by the Secretary's order, and it is proper for us now to determine the legal situation in regard to them when the temporary legislation expires. Under the permanent provisions of §§ 402 and 406 the Secretary's order was lawful, and the respondents present no grounds on which they can legally object to its application to them. The judgment of the Court of Appeals, setting the Secretary's order aside in part, must be

Reversed.