The petitioners, major domestic and foreign manufacturers of automobiles, have petitioned this Court for a review of an order of the National Highway-Traffic Safety Administration of the Department of Transportation, adopted pursuant to the National Traffic and Motor Vehicle Safety Act of 1966, 15 U. S.C. §§ 1381-1461, entitled “Motor Vehicle Safety Standard #208, Occupant Crash Protection in Passenger Cars, Multipurpose Passenger Vehicles, Trucks and Buses.” The Automobile Safety Act of 1966 was enacted as a response to the alarming number of deaths and injuries resulting from automobile accidents.1 Its expressed purpose is “ . . . to reduce traffic accidents and deaths and injuries to persons resulting from traffic accidents.” 15 U. S.C. § 1381. Chrysler Corp. v. Rhodes, 416 F.2d 319, 321 (1st Cir. 1969); General Motors Corp. v. Volpe, 321 F.Supp. 1112, 1115 (D.C.Del.1970). In achieving this goal, two courses of action are open to the Agency. (See Cong. Rep. No. 1919, 89th Cong.2d Sess., 1966, 2 U.S. Code, Cong. & Admin.News 2731 (1966).) It can act to prevent accidents,2 or it can act to prevent injuries in the event of accidents. Standard 208 is designed to accomplish the latter.
It is now established that most injuries caused by the impact of the automobile passenger with the steering wheel and column, the dashboard, the windshield, and other interior protrusions, can be prevented or at least ameliorated by safety-oriented vehicle design, and much attention has recently been devoted to the problem of the “second collision.”3
While many injuries of this sort can be prevented by the elimination from the interior surfaces of hard projections or sharp edges (e.g.: Standard 111, 49 C.F.R. § 571.111, S3.1.2.2) and by the use of energy absorbing steering columns (Standard 203, 49 C.F.R. 571.203), and by the application of energy absorbing materials to reduce impact forces at probable points of contact in the event of rapid deceleration, (Standard 201, 49 C.F.R. 571.201, S3.1, S3.4, S3.5), the most serious injuries can be prevented only by an occupant restraint device which absorbs the high deceleration forces while firmly preventing the passenger from being thrown against the inside of the vehicle or from being ejected out of it. The idea is to assure that when the car stops dead, the passengers don’t.
*664I
The standard under review requires the petitioners to build into their vehicles by a specified date a specified quantum of “passive protection” through the use of “passive restraint devices.” A passive restraint is defined as a protective occupant restraint device which does not depend for its effectiveness upon any action taken by the occupants beyond that necessary to operate the vehicle (36 F.R. 8296, May 4, 1971). An active restraint is a device which is not effective unless some action is taken by the occupants, the most familiar example of which is the fastening of a seat belt.
An “airbag” is a passive inflatable occupant restraint system. The term “airbag” is used generally to designate the entire system of apparatus in which a sensor, activated by the deceleration force of a collision, causes an explosive charge of compressed gas (or a gas generator) to rapidly inflate a large bag which restrains the occupant as he moves toward the windshield, dashboard or steering wheel of the car, and then deflates itself. This entire cycle, including the deflation, is completed in less than ' one-half second. Although the safety standard under review does not by its terms specify that airbags be used to meet the specified injury criteria, the petitioners unanimously contend that because the injury criteria of Standard 208 were established with the airbag in mind that the airbag is the only device which can be reasonably expected to satisfy these criteria, and that therefore, the standard is in reality an airbag requirement standard. Although nothing in the record justifies disagreement with the petitioners on this point,4 for the purposes of this opinion we do not find it necessary to distinguish between the airbag and any other form of passive restraint.
Standard 208 was first published as part of the initial federal standards issued pursuant to 15 U.S.C. § 1392(h) on February 3, 1967 (32 F.R. 2415 (1967) “Seat Belt Installation — Passenger Cars”) and established the requirements for seat belt installations. No objections were made to this requirement, and the standard remained unchanged until March 10, 1970. On that date the Agency published what is now generally referred to as Revised Standard 208, New Standard 208 or, more descriptively, the Airbag Standard. The amendment procedure has produced to date a series of twenty-four notices, consisting of notices of proposed rulemaking, notices of meetings, and various final amendments to the existing standard. Not all of these notices are important to the case presented to us, but a brief review of their chronology will help to put into perspective the course of the Agency’s action in the promulgation of Standard 208, and its reaction to the industry’s comments, and is necessary to an evaluation of many of the petitioners’ challenges to the procedure utilized by the Agency.
A proposed change in the initial Standard 208 was first announced in Notice 1, published on July 2, 1969, entitled “Inflatable Occupant Restraint Systems” (34 F.R. 11148). In this advance notice of proposed rulemaking, the Agency observed that a promising system of restraint, commonly referred to as “airbags,” was in the final stages of development and that it would be desirable that such a system be provided on new motor vehicles as soon as possible, and not later than January 1, 1972. Information was requested, and a public *665meeting was scheduled at which interested parties presented their views on the concept of mandatory, industry-wide installation of airbags (Notices 2 and 3, 34 F.R. 12107, 13480). Written comments were submitted as requested and as required by the APA § 4(c), 5 U.S.C. § 553(c).
After a lengthy evaluation of the comments of the manufacturers and other interested parties and newly developed technical information, the Agency issued a notice of a proposed safety standard entitled “Occupant Crash Protection; Passenger Cars, Multipurpose Passenger Vehicles, Trucks and Buses” (Notice 4, 35 F.R. 7187). The proposed rule generally provided a delay in the effective date of a passive restraint system from January 1, 1972 to January 1, 1973. However, during the interim period, passenger ear manufacturers were required to substitute a modified restraint system which amounted to an improved seat belt assembly. The proposal set out the procedures to be used in testing the devices and the injury criteria which must be met. A second public meeting was held (Notice 5, 35 F.R. 10120). The interim requirements of the proposed standard were modified to some extent by Notice 6 (35 F.R. 14941), which was published on September 25, 1970.
The Agency held numerous formal and informal meetings and conferences; comments were submitted by more than 120 interested persons. After consideration of these comments and other relevant materials, the Agency issued the passive protection requirements of the standard as a final rule in Notice 7 on November 3, 1970 (35 F.R. 16927). The effective date for passive systems was extended from January 1, 1973, to July 1, 1973, for front seat positions and to July 1, 1974, for all occupants; the injury criteria proposed in Notice 4 were modified in some respects. Notice 8, (35 F.R. 16937) issued simultaneously with Notice 7, was a proposal for further modifications of the standard as to a new requirement that deployable systems should not deploy in impacts of less than 15 miles per hour (mph), and for additional injury criteria for various test modes.
In response to a variety of objections raised in comments and in petitions for reconsideration, the Agency republished Standard 208 on March 10, 1971, as Notice 9 (36 F.R. 4600), a complete and final rule. This notice forms the basis of the Agency’s action which is before this Court for review. This notice incorporated a change in the effective date for passive restraints from July 1, 1973, to August 15, 1973, to reflect the historic model change-over period of the automobile industry. Certain concessions were made for open body vehicles and forward control vehicles, and the Agency indicated that it intended to issue further specifications for the anthropomorphic test devices which the manufacturers had suggested were too vague and which therefore produced inconsistent test data.
The Agency issued Notice 10 (36 F.R. 12858) in response to the petitions for reconsideration of Notice 9. This notice amended the requirements for the seat belt warning system option which was to become effective January 1, 1972. Notice 11 (36 F.R. 12866), issued simultaneously with Notice 10, proposed certain release requirements should a seat belt assembly be used as a passive restraint system, and requires that airbags be self-deflating.
Notice 12 (36 F.R. 19254) was issued on October 1, 1971, as a response to the petitions for reconsideration of Notice 9, and is the second of the two notices which this Court has been requested to review. This notice refers to a proposed rule, issued simultaneously, to delay one of the effective dates of Notice 9, and comments upon the necessity of later rulemaking in the standard. This amendment clarifies the monitoring system requirements for a passive system, specifies the positioning and locations of the anthropomorphic test devices and changes the cargo weight to be used for testing multipurpose passenger vehicles and trucks. In the comments to Notice 9, the *666Agency acknowledged that the test dummy specifications were inadequate and that variances in dummies could jeopardize the test results of a vehicle attempting to comply with the standard. The Agency stated that it would issue, at a later date, proposed amendments to the standard detailing performance and descriptive specifications for the test dummies. In the interim, the Agency stated in the comments to Notice 12 that if a vehicle is found to comply with the existing standard under a properly conducted test by a manufacturer the negative results of an Agency test will not be used as the basis for a finding of non-compliance so long as the difference in the test results can be attributed to the test dummies.
Notice 13 (36 F.R. 19266), issued simultaneously with Notice 12, proposed an amendment to Standard 208 which would allow for an additional interim option of a seat belt interlock system which will not allow the engine starting system in a vehicle to operate unless the driver and any front seat passengers have fastened their seat belts. Notice 14 (36 F.R. 19705) proposed a change to conform all explosive devices to existing state and federal regulations. Notice 15 (36 F.R. 23725) was a promulgation of several minor changes concerning seat belt warning systems. On February 24, 1972, the Agency published Notice 16, (37 F.R. 3911) as a final amendment to Standard 208, basically incorporating the proposals of Notice 13 which provided for the option of a manufacturer to use seat belts equipped with an ignition interlock system between August 15, 1973, and August 14, 1975. On March 16, 1972, in Notice 17 (37 F.R. 5507), the Agency proposed further amendments to the method of calculating the head injury criteria of Standard 208. Notices 18 to 24 (37 F.R. 10745, 12393, 13265, 16186, 16604, 22871, 23115) are not directly relevant to these proceedings except that portion of Notice 19 which adopted the proposed amendments of Notice 17.
Chrysler, Jeep, American Motors, Ford and the Automobile Importers of America have petitioned this Court for a review of Notice 9; Ford, American Motors and Jeep have petitioned this Court for a review of Notice 12. These petitions for review were consolidated in this Court, and were, argued together.
Notice 9, as amended by Notices 10, 12, 15 and 16 5 is designed to be implemented in three stages, the first two of which offer a manufacturer the choice of several options for compliance:
STAGE ONE: A manufacturer must provide on all vehicles manufactured between January 1, 1972, and August 14, 1973, one of the following three options:
(1) “Complete Passive Protection,” defined as a system which meets specified injury criteria6 for all seating *667positions in all impact modes, frontal (head-on into a fixed barrier at 30 mph), angular (30 degrees from either side of frontal into a fixed barrier), and side (90 degrees from frontal, impact at 20 mph with a lateral moving barrier) and which will prohibit any part of two test dummies from extending outside any part of the car in a rollover test (rollover in either lateral direction at 30 mph); or
(2) Sufficient interior padding plus lap belts such that the prescribed injury criteria are met at the front outboard positions with test dummies in a, 30 mph frontal crash into a fixed barrier; or
(3) Lap and shoulder belt systems with warning signals at the front outboard positions that restrain test dummies in a 30 mph frontal test crash without complete separation of the belts themselves or their anchorages, plus lap belts at other seating positions.7
STAGE TWO: A manufacturer must provide on all vehicles manufactured after August 15, 1973, one of the following three options:
(1) “Complete Passive Protection”; or
(2) Passive protection for front seat positions which will meet specified injury criteria in a 30 mph head-on crash into a fixed barrier, plus lap belts at all other positions; or
(3) Seat belts with ignition interlocks 8 for front seat passengers, plus non-interlocked belts at other positions which must meet the specified injury criteria for front outboard occupants in a 30 mph impact into a fixed barrier.
STAGE THREE: By August 15, 1975, a manufacturer must provide “Complete Passive Protection.”
II
A threshold issue is raised concerning this Court’s scope of review of Motor Vehicle Safety Standards. The manufacturers contend that this Court must determine whether the Standard is supported by “substantial evidence on the record as a whole,” and they rely upon the legislative history of the Act, upon the provisions of the Administrative Procedure Act, and upon Citizens to Preserve Overton Park, Inc. v. Volpe, 401 U.S. 402, 91 S.Ct. 814, 28 L.Ed.2d 136 (1971). The agency contends that because the rule under review is not the product of adjudication, but of informal rulemaking which is a legislative process, this Court’s review is restricted to a determination of whether the Agency complied with applicable procedural requirements and whether the Agency’s rule reflects a “rational consideration” of the relevant matters presented by interested parties.
We think that the scope of review urged upon this Court by the Agency is virtually no review at all; the Agency seems to ignore the existence of its record as well as the substantive limitations which the Act places upon its authority to set standards.9 Thus, even though the Agency may proceed by in*668formal rulemaking, the Act requires that several substantive criteria be met, and also provides for direct review of that rule by a United States Court of Appeals on the basis of the record which was compiled during the course of the rulemaking.
The Automobile Safety Act of 1966 (15 U.S.C. §§ 1381-1431) which empowers the Agency to set minimum performance standards for newly manufactured automobiles, provides that all standards “shall be practicable, shall meet the need for motor vehicle safety, and shall be stated in objective terms.” 15 U.S.C. § 1392(a). In addition, the Agency must consider relevant available motor vehicle safety data (15 U.S.C. § 1392(f)(1)), and must consider “whether any such proposed standard is reasonable, practicable and appropriate for the particular type of motor vehicle . . . for which it is prescribed.” 15 U.S.C. § 1392(f)(3). These factors represent the statutory minimum substantive criteria against which each automobile safety standard must be tested.
Congress established a distinct procedure for direct judicial review of Motor Vehicle Safety Standards by the United States Courts of Appeals. In such cases, the Secretary of Transportation is required to file in the Court the record of the proceedings “on which [he] based his order as provided in Section 2112 of Title 28 of the United States Code.” Section 2112 of Title 28 provides that “all of the evidence before the agency shall be included in the record,” excepting only that stipulated by the parties to be omitted as wholly immaterial to the questioned finding. Section 706(2) (E) of Title 5 provides that in a case in which statutory provision is made for review on the record of the agency, that the agency’s action must be supported by substantial evidence. Clearly when factual issues are involved, including the issue of whether compliance is technologically feasible, the reviewing court must consider the record upon which the Agency based its order.
The Agency contends that under the authority of California Citizens Band Assn. v. United States, 375 F.2d 43 (9th Cir.), cert. denied, 389 U.S. 844, 88 S.Ct. 96, 19 L.Ed.2d 112 (1967), it can act on the basis of evidence in its own files, and it does not have to make this evidence available to a litigant seeking judicial review. This may be true in a situation such as was present in that case, where the Agency needed only to make a discretionary decision unencumbered with factual or technical considerations of any kind. In that case, the petitioners requested a review of rules promulgated by the FCC dealing with the use of radio frequencies. The FCC had selected seven frequencies to be used by private radio operators under the new rules of interstate communications. There was no factual question raised in the selection of the frequencies; this was purely a matter of administrative discretion, and consequently was subject to a very narrow scope of review. Furthermore, the statutory language which empowered that Agency to act did not require the compiling of a record as the basis for its action. This is clearly not the case with the Automobile Safety Act which requires the Secretary to “consider relevant available motor vehicle safety data, including the results of research, development, testing and evaluation activities” and to “consult with the Vehicle Equipment Safety Commission, and such other State or interstate agencies (including legislative committees) as he deems appropriate.” 15 U.S.C. § 1392(f)(1) (2).
It seems clear that any rule which is required to be, inter alia, “practicable” and “objective” must be reviewed on the basis of the data which the Agency considered in its promulgation, regardless of whether formal hearings were or were not held. In Automotive Parts and Accessories Assn. v. Boyd, 132 U.S.App.D.C. 200, 407 F.2d 330 (1968), the Court was presented with the issue of whether informal rulemaking was authorized by the Act. The Court found that it was, and then upheld the safety standard *669there under review by looking at the underlying record to determine whether there was a factual basis for the Agency’s decisions, and whether the required “concise general statement” of the Agency’s purpose was properly supported.10 See also, Boating Industry Assn. v. Boyd, 409 F.2d 408 (7th Cir. 1969) where the identical procedure was followed.
An agency performing a rule-making function need not always compile a record of the material upon which its rule is based. It may act on the basis of data contained in its own files or on its own views or opinions. In such cases, a reviewing court cannot test the rule as promulgated against the evidence in the agency’s record. Flying Tiger Line, Inc. v. Boyd, 244 F.Supp. 889 (D.C.1965). In this case, however, the Agency’s freedom of action in regard to the rules it may promulgate is limited by a Congressional mandate which also requires the Secretary to compile a record. As observed in Overton Park, supra, the substantial evidence test may be applied to agency action even when the agency is performing a rulemaking function. 401 U.S. at 414-415, 91 S.Ct. 814, 28 L.Ed.2d 136. Accordingly, the reviewing court must “engage in a substantial inquiry” and a “thorough, probing, in-depth review,” yet at the same time must be mindful that the ultimate scope of its review is narrow and that it may not substitute its judgment for that of the agency’s. 401 U.S. at 416, 91 S.Ct. 814, 28 L.Ed.2d 136.11
Any safety standards issued pursuant to the Automobile Safety Act of 1966 must be tested for compliance with the statutory limitations of that Act, and this testing can only be done against the record. The record includes not only the notices of the Agency as published in the Federal Register and the comments and petitions for reconsideration submitted by interested parties, but also the technological and statistical data relied upon by the Agency in arriving at its conclusions. That this material is, or should be, a part of the record in this case is clear from the legislative direction that the Agency “consider relevant available motor vehicle safety data, including the results of research, development, testing and evaluation activities conducted pursuant to this chapter,” 15 U.S.C. § 1392(f)(1), and that the Secretary shall, in the event that a petition for review is presented to a Court of Appeals, file in that court the record upon which he based his actions. 15 U. S.C. § 1394(a)(1).12 To the extent that this conclusion is inconsistent with that *670reached by the Seventh Circuit in Boating Industry Assn. v. Boyd, supra, we respectfully decline to follow that decision.
Our review does not end with a determination that these statutory criteria have been met. Section 706(2) (A) of the Administrative Procedure Act requires a finding that the actual decision reached was not “arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law.” As pointed out in Overton Park, supra, to make this finding, the reviewing court must consider whether the decision was based on a consideration of the relevant factors and whether there has been a clear error of judgment. 401 U.S. at 416, 91 S.Ct. 814, 28 L.Ed.2d 136.
One further observation may serve to demonstrate the impracticality of the Agency’s argument that only the procedural aspects of a motor vehicle safety standard are subject to review brought pursuant to 15 U.S.C. § 1394. Once such a standard has been adopted and becomes effective, the validity of that standard is capable of being placed in issue in litigation in any of a number of ways. For example,13 15 U.S.C. § 1398 provides for the imposition of sanctions in the form of fines against anyone who manufactures for sale or offers for sale a motor vehicle which does not conform with applicable federal safety standards. An obvious defense in such a prosecution would be the claimed invalidity of the standard and it seems clear, that a previous judicial review limited to an approval of the procedural aspects of the adoption of the standard would not bar the defense. Such a defense might be successful at any judicial level, but the adjudication would almost certainly not be considered final until the Supreme Court had reviewed the issue or had refused to do so. In the interval preceding such final adjudication, which could be a substantial one, a chaotic condition would exist within the industry to the serious detriment of the manufacturers and the public alike, perhaps accompanied by inestimable damages to both. Such a possibility strongly militates against the limitation of review urged by the Agency and supports our determination that it should be here determined whether the standard is supported by “substantial evidence on the record as a whole.”14
*671In summary, the function of this Court is to test the “validity” of Standard 208. Although we are not empowered to substitute our judgment as to discretionary decisions made by the Agency, we will look to the underlying record to determine whether a factual basis exists for the Agency’s decisions and to determine whether the statutorily required concise general statement of the Agency’s purpose is properly supported. In order to be valid, Standard 208 must meet all statutorily prescribed criteria, and the Agency must have complied with all applicable procedural requirements. The former are contained in the Automobile Safety Act of 1966 and can be designated as substantive criteria, the latter, contained in the APA, can be designated as procedural requirements, and they can be discussed separately.
Ill
The petitioners’ first argument is that the Automobile Safety Act of 1966 does not authorize the Agency to establish a safety standard which requires the improvement of existing technology, and that the Agency may only establish performance requirements which can be met with devices which, at the time of the rulemaking, are developed to the point that they may be readily installed. The Agency’s response is that inasmuch as the technology of airbags is fully developed, and inasmuch as airbags are presently readily available to all manufacturers, this issue is not properly before the Court in this case. The Agency also contends that, even if that were not so, one of the prime purposes of the Act is to require automobile manufacturers to develop safety technology not presently existing.
The explicit purpose of the Act, as amplified in its legislative history, is to enable the Federal government to impel automobile manufacturers to develop and apply new technology to the task of improving the safety design of automobiles as readily as possible.15 The Senate Report, in a section entitled “Purpose and Need,” states:
“[T]his legislation reflects the faith that the restrained and responsible exercise of Federal authority can channel the creative energies and vast technology of the automobile industry into a vigorous and competitive effort to improve the safety of vehicles.” S.Rep. 1301, 89th Cong., 2d Sess., 2 U.S.Code, Cong. and Admin.News, 2709 (1966).
The same report continues:
“While the bill reported by the committee authorizes the Secretary to make grants or award contracts for research in certain cases, a principal aim is to encourage the auto industry itself to engage in greater auto safety *672and safety-related research.” Id. at 2718.
There is no suggestion in the Act that developed technology be in use by an automobile manufacturer or that any given procedure be an established industry practice prior to its incorporation into a federal motor vehicle safety standard. If the Agency were so limited, it would have little discretion to accomplish its primary mission of reducing the deaths and injuries resulting from highway accidents.
In fact, specific efforts by the Automobile Manufacturers Association to tie the rate of innovation imposed by safety standards to the pace of innovation of the manufacturers were rejected by the House Committee on Interstate and Foreign Commerce,16 and the reported bill proposed that safety standards be “practicable, meet the need for motor vehicle safety, and be stated in objective terms.”
In explaining the purpose of the first of these criteria, the House Report states:
“In establishing standards the Secretary must conform to the requirement that the standard is practicable. This would include consideration of all relevant factors, including technological ability to achieve the goal of a particular standard as well as consideration of economic factors.” H.R.Rep. 1776, p. 16.
If the Agency were limited to issuing standards only on the basis of devices already in existence, there would be no need for the Agency to give any consideration to the manufacturers’ technological ability to achieve a stated goal. Under this proposed interpretation, the Agency would be unable to require technological improvements of any kind unless manufacturers voluntarily made these improvements themselves. This is precisely the situation that existed prior to the passage of the Act, and we decline to eviscerate this important legislation by the adoption of this proposed in-terpretion. As it stands, the Act is reasonable, and the power of the Agency to “channel the creative energies and vast technology of the automobile industry into a vigorous and competitive effort to improve the safety of vehicles” fully meets the need for motor vehicle safety.
We do not intend to suggest that the Agency might impose standards so demanding as to require a manufacturer to perform the impossible, or impose standards so imperative as to put a manufae-*673turer out of business. But it is clear from the Act and its legislative history that the Agency may issue standards requiring future levels of motor vehicle performance which manufacturers could not meet unless they diverted more of their resources to producing additional safety technology than they might otherwise do. This distinction is one committed to the Agency’s discretion, and any hardships which might result from the adoption of a standard requiring, as does Standard 208, a great degree of developmental research, can be ameliorated by the Agency under 15 U.S.C. § 1392(c). This section allows the Secretary to extend the effective date beyond the usual statutory maximum of one year from the date of issuance, as he has done with Standard 208. The Senate report, in explaining this section, confirms the conclusions reached above:
“The power to specify a later effective date is needed because it may be a practical economic and engineering impossibility, as well as a source of great hardship and unnecessary additional cost, to require that all vehicle changes required by any new safety standard, whatever its scope or subject matter, be accomplished by all manufacturers on all their vehicles within 1 year.” S.Rep. 1301, 89th Cong. 2d Sess., 2 U.S.Code., Cong. and Admin.News, 2714 (1966).
Similarly, should it develop that, as the time for implementation of federally mandated devices approaches, such devices will require further testing or development, upon a showing to the Agency by the manufacturers or the developers the Agency should again review the matter and decide whether to extend the time for implementation or to alter, or even to abandon, the project.
In summary, the Agency is empowered to issue safety standards which require improvements in existing technology or which require the development of new technology, and it is not limited to issuing standards based solely on devices already fully developed. This is in accord with the Congressional mandate that “safety shall be the overriding consideration in the issuance of standards.” S.Rep. 1301, 2 U.S.Code, Cong. & Admin.News, p. 2714 (1966).
The petitioners next contend that Standard 208 is not practicable because airbag technology is not, at present, developed to the point where airbags can be installed in all presently manufactured cars. In light of our preceding conclusion, we need not discuss this contention at length. But we will observe that, as the record indicates, many of the development problems with which the petitioners have concerned themselves in their briefs (such as noise, sensor reliability, danger to out of position occupants and effectiveness in certain nonfrontal impact modes), have been eliminated or are presently the subject of continuing development efforts. We need not detail here the immense amount of factual data contained in the record relevant to this issue; suffice it to here observe that present systems demonstrate considerable sophistication over earlier prototypes. In addition, several automobile manufacturers and several airbag developers have expressed a great deal of confidence in their present systems and an equal confidence that present developmental research programs will eliminate any obstacles which may presently remain.
For example, General Motors announced in 1970 that final research and development efforts on its airbag system were currently underway, that it was entering the production design stage on those bags, and that it planned to introduce its system on a phased-in schedule so that by the fall of 1974, all 1975 model GM passenger cars and light trucks would have a passive protection system as standard equipment.17 Eaton has in*674dicated that it has largely eliminated noise as a problem and that it expected, in the near future, to solve the problem of the effect of airbag detonation on a child standing near the dashboard. On October 1, 1971, Irvin Industries stated that it had built a system which met or exceeded all published U.S. government specifications in effect at that time.18 Most recently, the Ford Motor Company delivered to the Chairman of the Board of Allstate Insurance Company an automobile equipped with an airbag restraint system for the front seat passenger. This vehicle was the first of a fleet of 200 similarly equipped cars which Allstate has purchased for use by selected employees in a joint field-test of these systems with Ford. While there is some disagreement as to the significance of this event, we do note that the insurance company involved has publicly expressed a large measure of confidence in the reliability and effectiveness of the system as delivered, and in the developmental research program of which this test is a vital part.
Since we have rejected the petitioners’ contention that nonexisting technology may not be the subject of motor vehicle safety standards, and in view of the present state of the art of passive inflatable occupant restraint systems, we conclude that Standard 208 is practicable as that term is used in this legislation.19
The petitioners contend that Standard 208 does not meet the need for motor vehicle safety because belts offer better protection to occupants than do airbags. The Agency defends the standard by contending that airbags offer better protection to occupants than do belts. The record. supports the conclusion that each type of occupant restraint offers protection in a slightly different form for differing impact situations. Neither is clearly superior to the other in every respect. Consequently, we conclude that the Agency’s decision to abandon active restraints in favor of passive restraints was a proper exercise of its administrative discretion.
Paramount among the Agency’s considerations in deciding to require all occupant restraint systems to be fully passive was the factor of low belt usage. It is uncontested that active restraints are not extensively used. The record indicates that usage rates for lap belts are about 20 to 30%, and for the lap and shoulder harness combination about 1 to 5%; it is projected that devices (or laws) to encourage or to force belt usage will not increase usage rates above 60%. The petitioners’ position is that, if the Agency starts with the proposition that occupants are not now using belts, its consequent course of action should be an effort to increase belt usage through whatever means are available (e. g.: ignition interlocks, compulsory usage laws) rather than to disregard active restraints entirely and require passive protection in all vehicles.
It is conceded that belts, when used, are extremely effective. The conclusive evidence on this point is a study of more than 28,000 accident cases in Sweden which showed that no occupant wearing a combined lap belt and shoulder harness was fatally injured in any accident occurring at speeds below 60 miles per hour.20 On the other hand, while belts *675are superior to airbags in some respects, most notably in rollovers and multiple impact situations, airbags have advantages over belts in other equally important respects. For example, airbags spread crash deceleration forces over most of the whole body, while belts concentrate them on the narrow area of the rigid belt. Airbags restrain the body evenly over a greater distance and a longer period of time by permitting occupants who are thrown forward in the crash to ride down the deceleration forces more gradually, over a distance of two or three feet. An airbag system, being passive, removes the elements of the occupants’ will, memory and skill from the consideration of reliability. Furthermore, many people cannot properly use belt systems (e. g.: children under four years of age, persons under 55 inches in height, obese or very smallwaisted persons and pregnant women) and among those who can, there is an inevitable percentage who will not wear them properly. We cite these facts not to advocate belts over airbags or vice versa, but to indicate the myriad factors which must be carefully considered by the Agency in reaching a conclusion on this issue.
We conclude that the issue of the relative effectiveness of active as opposed to passive restraints is one which has been duly delegated to the Agency, with its expertise, to make; we find that the Agency’s decision to require passive restraints is supported by substantial evidence, and we cannot say on the basis of the record before us that this decision does not meet the need for motor vehicle safety.
IV
We now turn to the final major substantive argument presented by the petitioners: that Standard 208 fails to meet the statutorily required criteria of objectivity. The necessity for objective certainty in the performance requirements of safety standards was clearly recognized by Congress in the Safety Act. The Act provides, as noted above, that “standard [s] shall be practicable, shall meet the need for motor vehicle safety, and shall be stated in objective terms.” 15 U.S.C. § 1392(a). These requirements are repeated in the statutory definition of motor vehicle safety standards : “ ‘Motor vehicle safety standards’ means a minimum standard for motor vehicle performance, or motor vehicle equipment performance, which is practicable, which meets the need for motor vehicle safety and which provides objective criteria.” 15 U.S.C. § 1391(2). The requirement of objectivity was emphasized in the following statement from the House Report: “In order to insure that the question of whether there is compliance with the standard can be answered by objective measurements and without recourse to any subjective determination, every standard must be stated in objective terms.” H.R. 1776, 89th Cong. 2d Sess.1966, p. 16.
The importance of objectivity in safety standards cannot be overemphasized. The Act puts the burden upon the manufacturer to assure that his vehicles comply under pain of substantial penalties.21 In the absence of objectively defined performance requirements and test procedures, a manufacturer has no assurance that his own test results will be duplicated in tests conducted by the Agency. Accordingly, such objective criteria are absolutely necessary so that “the question of whether there is compliance with the standard can be answered by objective measurement and without recourse to any subjective determination.”
*676Objective, in the context of this case, means that tests to determine compliance must be capable of producing identical results when test conditions are exactly duplicated, that they be decisively demonstrable by performing a rational test procedure, and that compliance is based upon the readings obtained from measuring instruments as opposed to the subjective opinions of human beings.22 Standard 208 requires that compliance be determined by specified tests using an anthropomorphic dummy built to the specifications of SAE Recommended Practice J963, “Anthropomorphic Test Device for Dynamic Testing.” These specifications generally provide for the structural characteristics of the dummy, which is to simulate the basic human body components in size, shape, mass and kinematics. The petitioners contend that this test device will not produce consistent, reliable, or repeatable test results.
The record supports the conclusions that the test procedures and the test device specified by Standard 208 are not objective in at least the following respects: (1) The absence of an adequate flexibility criteria for the dummy’s neck; the existing specifications permit the neck to be very stiff, or very flexible, or somewhere inbetween, significantly affecting the resultant forces measured on the dummy’s head. (2) Permissible variations in the test procedure for determining thorax dynamic spring rate (force deflection characteristics of the dummy’s chest) permit considerable latitude in chest construction which could produce wide variations in maximum chest deceleration between two different dummies, each of which meets the literal requirements of SAE J963. (3) The absence of specific, objective specifications for construction of the dummy’s head permits significant variation in forces imparted to the accelerometer by which performance is to be measured.23
The shortcomings of the test device become understandable, if still not excusable, when one considers the minutes of a May 23, 1971, meeting of the SAE Crash Test Dummy Subcommittee, which are a part of the record in this case. That report points out that since SAE J963 only relates some limited performance specifications for an anthropomorphic test device, it should be used as a guide for specifications of such a device, with considerations given to certain enumerated shortcomings. Further, the report mentions that “[i]t should be noted that the original document [SAE J963] was intended to specify a research tool and, therefore, the document was written in general terms,” and that until more definitive specifications are published, “[t]he subcommittee is of the opinion that SAE Recommended Practice J963 can only be used as a guide rather than as a rigid set of specifications.” The Safety Systems Laboratory (formerly of the office of Vehicle Systems Research, National Bureau of Standards, now of the NHTSA) reported in September of 1971 that SAE J963 is not “sufficiently definitive to provide an adequate descrip*677tion of the dummies” and concluded that “further study is necessary to develop standard specifications for an anthropometric crash dummy.” A similar determination was made by the Cornell ■Aeronautical Laboratory, which concluded: “More complete specification of anthropomorphic dummy characteristics is needed. While it is tentatively concluded that a degree of repeatability is obtainable with existing dummies, the dynamic performance of these devices needs specification designed to insure repeatability.”
Further recitation of statements from the record that the test procedure and devices of Standard 208 lack objectivity is unnecessary because, interestingly enough, the Agency has never asserted that its test criteria are fully objective. Although the Agency noted, at the time it first proposed the adoption of SAE J963, that these dummy specifications “may not provide totally reproducible results,” it justified their adoption because they are “evidently the most complete set available at this time.” 35 F. R. 7188. Specific comments were requested from interested parties; comments were submitted by each of the parties to this action.24 The Agency replied to these comments in Notice 7 by modifying the specifications slightly, but retained the requirement of the SAE J963 dummy, noting only that “dummies conforming to the SAE specifications are the most complete and satisfactory ones presently available.” 35 F.R. 16928. In Notice 9, the Agency again remarked not that the SAE J963 dummy was adequate, but that it was the “best available,” and that it was sponsoring research in this area and would therefore issue new specifications at some time in the future. Virtually all automobile manufacturers, including each of the present petitioners, submitted petitions for reconsideration to these requirements of Notice 9. General Motors went so far as to supply suggested modifications and additions to the SAE J963 test device specifications which it felt were necessary before a device would provide consistent test results.
Finally, the Agency conceded in its most recent official pronouncement on the subject, Notice 12, that the specifications contained in SAE J963 “do not completely define all the characteristics of the dummies that may be relevant to their (and the vehicle’s) performance in a crash test.” 36 F.R. 19255. That notice, however, retains the requirement.
The Agency makes two responses to the attacks upon the standard’s lack of objectivity. First, it contends that the incompleteness of the dummy device is not fatal to the standard because it has assured the manufacturers that specifications for a suitable test device will be issued in the future.25 We do not think it necessary to dwell upon the obvious inadequacy of this response because it seems to us axiomatic that a manufacturer cannot be required to develop an effective restraint device in the absence of an effective testing device which will *678assure uniform, repeatable and consistent test results.
The Agency’s second response is that it has assured the manufacturers that their position with regard to compliance will not be jeopardized by the variances in the test device. The Agency’s solution to the problem was set out in Notice 12:
“If the NHTSA concludes after investigation that a manufacturer’s tests are properly conducted, with dummies meeting the specifications, and show compliance with the standard, and that differences in results from tests conducted by the agency are due to differences in the test dummies used by each, the Agency tests will not be considered to be the basis for a finding of noncompliance.” 36 F.R. 19255.
This statement is illusory. This test allows as many as three subjective judgments to be made by the Agency, including the ultimate determination of whether the differences in results from tests conducted by the Agency are due to differences in the test dummies used by each. As noted above, objectivity requires that each essential element of compliance be made by specified measuring instruments; there is no room for an “agency investigation” in this procedure. The inherent uncertainty of this procedure is no substitute for the specification of an adequate test device, especially in light of the Agency’s power to specify such a device. As the Agency noted in Notice 12:
“[S]ince the dummy is merely a test instrument and not an item of regulated equipment, it is not necessary to describe it in performance terms; its design could legally be ‘frozen’ by detailed, blueprint-type drawings and complete equipment specifications.” 36 F.R. 19255.
While we have concluded that automobile manufacturers can (and should) be compelled by automobile safety standards to develop new safety devices, we hold that the performance goals which they must meet must be clearly delineated by the Agency. That is to say, while they can be required to develop a new device not presently existing, we do not think that they can (or should) simultaneously be required to develop a testing device by which the safety device is to be measured.
The Amicus Center for Auto Safety cautions that if we were to hold that the standard is invalid because the test procedures are not objective we would be allowing “the tail to wag the dog.” We think that the Center has misapprehended the anatomy of the animal. The Act requires the Agency to issue performance standards, which a manufacturer must then meet by any system of hardware it chooses. It is clear from a reading of the Act and its legislative history that the performance standards (injury criteria and test procedures) are in fact the “dog” to which a manufacturer is free to attach the tail of its choosing. To rule otherwise would be to permit the Agency to establish a “product standard” requiring airbags, without requiring it to establish adequate, objective and repeatable test procedures by which a manufacturer’s airbags can be measured, by it or by the Agency, for compliance.
Since we find that the dummy as specified is inadequate for the testing purposes for which it is designated, we do not find it necessary to pass upon the petitioners’ contention that the dummy is a poor predictor of human response; nor do we reach their contention that the standard is unconstitutionally vague.
V
The petitioner Automobile Importers of America26 contends that *679Standard 208 effectively eliminates convertibles and sports cars from production because these types of cars are inherently incapable of meeting the injury criteria of the standard. At issue is the question of whether the standard violates 15 U.S.C. § 1392(f)(3) which provides that “In prescribing standards under this section, the Secretary shall . . . consider whether any such proposed standard is . . . appropriate for the particular type of motor vehicle . . . for which it is prescribed.” The implications of this section that safety standards shall take into consideration different classes of vehicles, such as small cars, sports cars, or convertibles, and that standards shall not be used as a device to bring about their extinction, is made clear from the Act’s legislative history:
“[I]n determining whether any proposed standard is ‘appropriate’ for the particular type of motor-vehicle equipment or item of motor-vehicle equipment for which it is prescribed, the committee intends that the Secretary will consider the desirability of affording consumers continued wide range of choices in the selection of motor vehicles. Thus it is not intended that standards will be set which will eliminate or necessarily be the same for small cars or such widely accepted models as convertibles and sports cars, so long as all motor vehicles meet basic minimum standards. Such differences, of course, would be based on the type of vehicle rather than its place of origin or any special circumstances of its manufacturer.” S.Rept. 1301, 2 U.S.Code, Cong. & Admin.News, 2714 (1966).
Congressman Springer observed in the House debate that “ [i] t is not the intention to deprive the public of cars by outlawing types and models. . . . [This legislation] avoids imposing on the buying public a complete standardization to humdrum, dreary vehicles.” 112 Cong.Rec. 19630-31 (1966). The Act clearly contemplates the continued production of sports cars and convertibles.
The special difficulties facing soft top convertibles and sports cars become obvious upon reflection. For example, Standard 208 requires a rollover test throughout which all portions of the test dummy must remain contained within the outer surfaces of the vehicle passenger compartment (j[ S 5.3 and ff S 6.-1). The comments submitted to the Agency by virtually every automobile manufacturer indicate that a considerable difficulty is being encountered with this requiremnt for vehicles with fixed steel tops because of the tendency of the roof to rupture or be punctured which, when combined with the centrifugal force operating on the dumipy during the rollover, will allow the hands and arms to raise over the dummy’s head and extend beyond the vehicle body perimeter.
While sufficient reinforcement should solve the problem currently being experienced with the hard top vehicles, it is obvious that a soft top convertible is inherently incapable of passing such a rigid requirement. It seems equally obvious that while the purchaser of a fixed roof automobile should expect some measure of protection from that roof, (perhaps expecting more than he is currently obtaining) owners of convertibles would not be disappointed to find that they are being afforded less protection in this area. Other problems inherent to convertibles and sports cars have been brought to the Agency’s attention by several automobile manufacturers, do*680mestic and foreign, but little if any agency action has been taken in response to these comments. This lack of response is surprising in view of the Agency’s obligation to afford such vehicles special consideration, as detailed above.
Accordingly, this issue is remanded to the Agency for further consideration without prejudice to the petitioner’s right to resubmit this issue to this Court for review after the Agency has been afforded an opportunity to amend Standard 208 so that it does not in fact serve to eliminate convertibles and sports cars from the United States new car market.
VI
The petitioners’ principal procedural arguments can be consolidated into the assertion that the Agency has acted arbitrarily, capriciously, irresponsibly and unlawfully by requiring manufacturers to comply with Standard 208 by a fixed date while it has consistently been issuing a rapid succession of final orders, each of which has been incomplete in several significant respects, (several have been coupled with proposed rulemaking to fill in missing parts), thereby constantly changing the basic requirements of the standard (performance criteria and test procedures), and more proposed changes are still pending. The Agency’s response is that its actions in this regard reflect the extreme flexibility which is the very purpose of delegated rulemaking authority. We agree. The Safety Act gives the Secretary the authority “to issue, amend, and revoke such rules and regulations as he deems necessary to carry out [the Act].” 15 U.S.C. § 1407. This broad delegation of rulemaking authority allows the Agency to amend its safety standards in those complex and technical respects which are brought to its attention as a result of petitions for reconsideration filed by “interested persons.” 27 In this way the Act contemplates that a safety standard might evolve through a series of industry-suggested and agency-adopted amendments, as has been done in this case. This procedure is consistent with one of the central purposes of the Act that the automobile industry be encouraged to develop new and improved safety technology.
It is equally important to note that almost all of the changes in the requirements of which the petitioners complain were made as a result of comments or petitions for reconsideration submitted to the Agency by the petitioners themselves, and these amendments to the standard generally have relaxed the requirements or have extended the date of implementátion to provide the manufacturers with additional leadtime, or to conform the date of implementation with the traditional date of the annual model changeover. These petitioners cannot be heard to complain that the Agency has constantly revised its standard when those changes were initiated by these very petitioners. We find no irregularities in the procedure utilized by the Agency in this regard.
Finally, the petitioners contend that the Agency has acted unlawfully in compiling the record which has been submitted to this Court. Specifically, they contend that substantial relevant materials were not placed on the docket of the proceeding or otherwise made available to interested parties until after petitions for review were filed in this Court. The petitioners’ position is that this has prejudiced them by depriving them of an opportunity to comment upon these materials prior to the issuance of the *681standard in that this procedure violated the Agency’s expressed policy of placing “all technical material, not routinely available through other sources, supporting or otherwise relating to proposed safety standards” in the record. (Notice 4, 35 F.R. 7188.)
We have examined the materials of which the petitioners complain, and we find that the late submission of this material (if in fact it was not timely submitted), did not serve to prejudice the petitioners in this case. The materials to which the petitioners refer consist principally of monthly progress reports from ongoing research projects conducted under the direction of the Agency in accordance with 15 U.S.C. § 1395. Prior to the filing of the petitions for review in this case, the Agency placed in its general reference section of its file a research input memorandum which contained a list of the projects then in progress.28 After these materials were docketed, several petitioners submitted comments to the Agency in response to these materials. These comments were therefore before the Agency when it issued Notice 12, and the petitioners have not alleged that the Agency failed to consider these comments in the issuance of this notice. Therefore, in that the petitioners were not prejudiced by the Agency’s manner of compiling the record,29 we do not reach the issue of whether substantial irregularities did in fact occur in the compiling of the record which is before this Court. Similarly, we do not reach a determination of whether or not the Agency’s announced policy rises to the level of a “procedural regulation” as is contended by the petitioners, although we express some doubt in this regard.
VII
We have considered the petitioners’ remaining arguments and find either that they have been effectively answered by the conclusions reached above or that they are without merit.
We conclude that Paragraph S 8.1.8 of Standard 208 which requires the use of the anthropomorphic test device as defined in SAE Recommended Practice J963 to be invalid; the remaining portions of the standard which do not depend for their effectiveness upon Paragraph S 8.1.8 are valid and remain in effect. 49 C.F.R. § 571.9. The proceeding is remanded to the Agency with instructions that any further specifications for test devices be made in objective terms which will assure comparable results among testing agencies, and that the effective date for the implementation of passive restraints be delayed until a reasonable time after such test specifications are issued.
. “Impelled by the awesome destruction of human lives and property on the highways and concerned over serious allegations from many quarters that some automobiles presented serious hazards to safety, Congress authorized [this legislation].” General Motors v. Volpe, 321 F.Supp. 1112, 1115 (D.C.Del.1970) (Footnote omitted.)
. E. g.: A new safety standard has recently been proposed (37 F.R. 7210, April 12, 1972) which, if adopted, will set standards for, inter alia, maximum allowable obstructions in the driver’s fields of direct view, light transmittance levels of glazing materials that provide fields of direct view, and visibility of the corners of the vehicle. The express purpose of this rule is to “reduce the likelihood that a motor vehicle will collide with a pedestrian, stationary object, or another vehicle because the driver either did not see the object collided with or saw it too late to avoid collision.”
. The Senate Commerce Committee expressed its concern this way:
“For too many years, the public’s proper concern over the safe driving habits and capacity of the driver (the ‘nut behind the wheel’) was permitted to overshadow the role of the car itself. The ‘second collision’ — the impact of the individual within the vehicle against the steering wheel, dashboard, windshield, etc. — has been largely neglected. The committee was greatly impressed by the critical distinction between the causes of the accident itself and causes of the resulting death or injury.”
S.Rept. No. 1301, 89th Cong., 2d Sess., 1966, 3, 2, U.S.Code, Cong. and Adm. News, 2710 (1966).
. This is not to imply that the Agency is prohibited from issuing this type of standard, in certain instances, if it wishes to. For example, the Agency might establish minimum performance requirements for braking systems which the industry could only meet by the use of braking systems not necessarily in broad current use. Even though such a standard would in effect require such devices, the industry would not be foreclosed from developing and implementing other systems which could also comply. We do not hold that the Agency is precluded from such a course of action; we only conclude that it has done so in this case.
. None of the consolidated cases request a review of Notices 10, 15, 16, or 19 to 24. Notices 10 and 15 constitute minor changes in the existing standard. Petitions for review have recently been filed in this Court by some of the present petitioners requesting a review of Notice 16 and a determination of its validity. That case has not yet been briefed or argued to this Court. We do not, therefore, discuss in this opinion the relative merits of Notice 16 or rule upon its validity. However, for the purposes of this opinion, we must presume it to be valid. Pacific States Box & Basket Co. v. White, 296 U.S. 176, 185, 56 S.Ct. 159, 80 L.Ed. 138 (1935).
. The injury criteria specified by Standard 208 are maximum values that may be read by instruments placed in the head, chest and upper legs of instrumented test dummies. In the case of chest impacts, the applicable criterion is expressed in terms of “g’s,” a measurement of deceleration. Except for “peaks” of very short duration, the chest’s deceleration may not exceed 60 g’s. For the head, the criterion is expressed as a “severity index,” a mathematically computed value which is intended to take into account not only “g” levels but also the total duration of the numerous “peaks” recorded by the instruments ; the maximum severity index permitted is 1,000, as calculated by the method adopted in Notice 19. In the case of the knee and upper leg, the criterion is load (expressed in pounds), and a maximum of 1,400 pounds is allowed.
. Under options 2 and 3 the outboard positions, front and rear, must have emergency locking or automatic retractors for lap belts. Shoulder belts can have manual adjustment or emergency locking retractors. Belts must release at a single point by pushbutton action. All belt warning systems must have a continuous or intermittent audible signal and a continuous or flashing light displaying “Fasten Seat Belts” or “Fasten Belts” when the ignition switch is on and the transmission gear selector is in any forward position.
. The interlock system must prevent the starting of the engine if any front seat occupant does not have his belt fastened. The occupant must be seated before the belt system is fastened, and unfastening a belt after the engine is started may not stop the engine, but must activate a light-sound warning system.
. The Amicus Center for Auto Safety considers these limitations to be “only the most general of constraints.” This is a contention, in effect, that the Standard itself cannot be reviewed, bht only the procedure from which it evolved.
. The Court found that there was substantial support in the record for the conclusion that the safety device which was the subject of the standard under review contributed to consumer safety so as to warrant its inclusion in all newly manufactured automobiles. 407 F.2d at 342. The Court observed that its function as a reviewing court was to determine whether the Agency acted in a “manner calculated to negate the dangers of arbitrariness and irrationality in the formulation of rules for general application.” 407 F.2d at 338. We fail to see how a court can make this determination without reviewing the record upon which the Agency’s rule was based.
. In 1904, Mr. Justice Brown summarized these principles in Bates & Guild Co. v. Payne, 194 U.S. 106, 109-110, 24 S.Ct. 595, 597, 48 L.Ed. 894 (1904) :
“[W]here the decision of questions of fact is committed by Congress to the judgment and discretion of the head of a department, his decision thereon is conclusive ; and that even upon mixed questions of law and fact, or of law alone, his action will carry with it a strong presumption of its correctness, and the courts will not ordinarily review it, although they may have the power, and will occasionally exercise the right of so doing.”
See also, L. Jaffe, Judicial Control of Administrative Action, 575-76 (1965).
. While not necessary to a determination of this question, we note that the Congressional Committee reports are explicit on this subject. The House report states that whether the Secretary proceeds formally or informally he must “establish a record which shall be the basis for his actions” and that the review provision of Section 105(a)(3) of the Act incorporating Section 10 of the APA “means (1) that a reviewing court will consider the entire rec*670ord before it and (2) that the findings of the Secretary will be sustained when supported by substantial evidence on the basis of the entire record.” H.R.Rep.No.1776, 89th Cong., 2d Sess., pp. 16-21. The Senate report is to the same effect, S.Rep. No.1301, 89th Cong., 2d Sess., pp. 7-8.
. The defense that the standard is invalid because it is not practicable, does not meet the need for motor vehicle safety, or is not stated in objective terms might also be raised in an action brought pursuant to 15 U.S.C. § 1399. This section confers jurisdiction upon the United States District Courts to restrain violations of the Act or to restrain the sale or the introduction into interstate commerce of any motor vehicle or item of motor vehicle equipment which is determined not to conform to applicable Federal standards. Also, we note that 15 U.S.C. § 1397(c) expressly preserves all common law liability with respect to compliance with any Federal motor vehicle safety standard. Therefore, although compliance with Federal standards is not a defense to a products liability action for failure to use reasonable care in design or construction to avoid subjecting the user to an unreasonable risk of injury, non-compliance with a valid safety standard may constitute negligence per se. See: Larsen v. General Motors Corp., 391 F.2d 495, 503, n. 5 (8th Cir. 1968).
. The legislative history of the Act supports this conclusion. (See footnote 4, supra.)
“Any person who believes himself to be adversely affected by the promulgation of a standard may obtain judicial review, in accordance with section 10 of the Administrative Procedure Act. The Administrative Procedure Act sets forth the long-established criteria for judicial review of agency action and provides that agency findings shall be upheld if supported by substantial evidence on the record considered as a whole.” Senate Report No. 1301, .Tune 23, 1966, 2 U.S. Code, Cong. & Admin.News, 89th Cong., 2d Sess., 2709, 2715-16 (1966).
In addition, we note that Professor Davis, after an examination of the various pro*671posed scopes of review of evidence concludes that “review of evidence, as distinguished from review of other questions in a ease, probably cannot feasibly be reduced to something less than what is customary under the substantial-evidence rule.” 4 O. Davis, Administrative Law Treatise, § 29.07 (1958).
. This is not to imply that the automobile manufacturers were intended to bear the entire burden of developing the required technology. To the contrary, the Act established a research and development section in the Agency itself (15 U.S.C. § 1395), and it is clear that automobile equipment manufacturers have a substantial interest in and are also encouraged to develop new safety devices, for their own profit, if for no other reason. For example, none of the petitioners in this Court is developing its own airbag system. Presently, systems are being developed by General Motors, which has announced that its system will be available, in whole or in part, to any other manufacturers who wish to purchase it, and by Eaton, Inc., Allied Chemical Corp., and Irvin Industries. The process of innovation envisioned by Congress was concisely outlined by Congressman Springer:
“Then, as new and better systems and devices are developed, either by the automobile industry itself, the equipment manufacturers, or by research and experiment supported by private or governmental sources, those devices will become part of the expected standard for late models.” 112 Cong.Rec. 19630 (1966).
. The Automobile Manufacturers Association transmitted to the House Committee several amendments to HR 13228 (the House Bill) proposing a requirement that:
“ . . . the Secretary, in proposing and issuing orders establishing, amending, or withdrawing Federal motor vehicle safety standards under this section, shall be guided so far as practicable by the following criteria, and the Secretary shall include in each such order findings of fact with respect thereto :
“(2) The standard shall be consistent with the continuation or adoption by motor vehicle manufacturers of efficient designing, engineering, and manufacturing practices, and with innovation, progressiveness, and customary model changes in the automotive industry.
“(3) The standard, the means of complying with the standard, and the methods of testing for compliance should embody feasible devices and techniques that are available or can be made available in a reasonable time and at costs commensurate with the benefit to be achieved.
“(5) The standard should be made effective so as to allow adequate time for compliance, taking into account the time required for designing, engineering, tooling and production.
Hearings Before the Committee on Interstate and Foreign Commerce, U.S. House of Representatives, 89th Cong., 2d Sess., on H.R. 13228, “Part 2, Traffic Safety,” p. 1203.
None of these specific restraints sought by the Automobile Manufacturers Association was adopted, and we must decline to write into the Act the very same suggestions which Congress declined to write into the Act.
. This schedule was announced in June, 1970; on February 25, 1972, General Motors revised its schedule somewhat and now plans to introduce an experimental system for the front seat occupants in up to 1,000 1973 model cars.
. These statements refer to the requirements of the standard in effect at that time, which required passive protection for front seat occupants in frontal crashes only for a deadline of August, 1973. Eaton, on July 8, 1971, indicated that it was prepared to produce airbag systems which it felt could meet these requirements for all cars of all sizes and that, if orders were made immediately, it could produce such systems for the 1973 deadline.
. The House Report indicates that “practicable” requires consideration of all relevant factors, including the technological ability to achieve the goal of a particular standard. See supra, at page 19.
. Another analysis of 160 accident cases revealed that 99% of all lap and shoulder belt users “either had no injury or only minor injury,” and that the only two fatalities among such persons occurred *675“under unusual circumstances.” Xelson, Lap-Shoulder Restraint Effectiveness in the United States, Automotive Engineering Congress (1971).
. The Act authorizes civil penalties of up to $1,000 for each nonconforming vehicle, subject to a maximum of $400,000. 15 U.S.C. § 1398(a). Injunctive relief is also authorized to restrain the sale or the importation into the United States of any motor vehicle or item of motor vehicle equipment which is determined not to conform to applicable Federal motor vehicle safety standards. 15 U.S.C. § 1399 (a).
. This definition of objectivity is taken from the Amicus Center for Auto Safety’s comments to the publication of Notice 4, in which the Center objected to the lack of objectivity inherent in the specified test device. With reference to the objections voiced in part II of the dissent, it is observed that a test procedure such as the rollover test, which is dependent upon simple visual observation, where there is no room for disagreement concerning the results and which is not dependent upon the subjective opinions of human beings would be objective as that term is used in this legislation.
. Section 4.1 of SAE J963 states only that: “The head shall consist of composite structures that are geometrically similar to the human head. The basic structure shall have an accessible internal ballast and instrumentation cavity and a pliable external covering with appropriate surface contours.” The absence of any specification for the thickness and impact characteristics of the pliable external covering of the dummy may also result in significant variations in forces recorded on the dummy during an impact.
. The Amicus Center for Auto Safety, in its comments to the Agency concerning Notice 4, concluded:
“SAE J963 is actually an incomplete specification for a design. Numerous motion characteristics as functions of forces are not mentioned. . . . This means that any person desiring to acquire a dummy must hire an engineer to complete the design. It also means no two auto manufacturers would be likely to test with identical configurations of dummies unless they obtained their dummies from the same source.”
. The Agency stated in Notice 12:
“Considerable development work is in process under various auspices to refine the dynamic characteristics of anthropomorphic devices, to determine which designs are most practicable, offer the most useful results, and best simulate the critical characteristics of the human body. The NHTSA is monitoring this work (and sponsoring some of it), and intends to propose amendments of the standard in accordance with it to add more detailed performance and descriptive specifications for the test dummies, although no changes are being made in that respect by this notice.” 36 F.R. 19255.
. None of the other petitioners in this case make this argument because none of the other petitioners make convertibles or sports cars in any significant numbers. These types of automobiles, however, constitute a major portion of the production of this petitioner. The Automobile Importers of America is an unincorporated *679association representing its member companies: Alfa Romeo, Inc., British Ley-land Motors, Inc., Bayerische Motoren Werke AG, Citroen Cars Corporation, Van Doome’s Automobielfabrieken N.Y., FIAT Motor Company, Inc., Fuji Heavy Industries Ltd., American Honda Motor Co., Inc., Lotus Cars Ltd., Mazda Motors of America, Inc., Mitsubishi Motors Corporation, Nissan Motor Corp. in U.S.A., Peugeot, Inc., Renault, Inc., Rolls-Royce, Inc., SAAB-SCANIA of America, Inc., Toyota Motor Sales, U.S.A., Inc., and Volvo, Inc.
. The Agency has promulgated regulations which permit any interested person to file petitions for reconsideration, 49 C.F.R. § 558.85, and which allow the Secretary either to “issue a final decision on reconsideration without further proceedings, or he may provide such opportunity to submit comment or information and data as he deems appropriate,” 49 C.F.R. § 553.37, and to “initiate any further rulemaking proceedings that he finds necessary or desirable.” 49 C.F.R. § 553.25.
. The Agency’s technical reference section also maintained a list of current research projects.
. This is not to understate the importance of the record to this case. Our comments in section II of this opinion clearly indicate that the entire record upon which the Secretary based his order is essential to a subsequent judicial review of his actions.