For information about air bags and related rulemakings: Visit the NHTSA web site at http://www.nhtsa.dot.gov and select "AIR BAGS: Information about air bags."

For non-legal issues: Mr. Clarke Harper, Chief, Light Duty Vehicle Division, NPS-11, National Highway Traffic Safety Administration, 400 Seventh Street, SW, Washington, DC 20590. Telephone: (202) 366-2264. Fax: (202) 366-4329.

For legal issues: J. Edward Glancy, Office of Chief Counsel, NCC-20, National Highway Traffic Safety Administration, 400 Seventh Street, SW, Washington, DC 20590. Telephone: (202) 366-2992. Fax: (202) 366-3820.

SUPPLEMENTARY INFORMATION:

I. Background.

A. How Air Bags Work.
B. Circumstances of Air Bag Fatalities.

II. The Safety Problem: Frontal Impacts and Air Bags--Lives Saved, and Lives Lost.

A. Frontal Impacts.
B. Air Bags: Lives Saved, and Lives Lost.

III. Search for Solutions.

A. The Early Years.
B. The Last Five Years.
C. Recent Petitions for Rulemaking.

IV. Overview of Comprehensive NHTSA Plan for Addressing Problem.

V. Depowering Air Bags

A. Results of NHTSA Test Program
B. Effects of Depowering and Optimizing

1. Passenger Air Bags
   
2. Driver Air Bags

C. Alternative Proposals

1. Approach I--Temporary Change in Unbelted Chest Acceleration Requirement.
   
2. Approach II--Temporary Replacement of Unbelted Crash Test Requirement with a Sled Test Protocol Incorporating a Standardized Crash Pulse.
   
3. Request for Additional Information.

D. Consideration of Other Alternatives.
E. Effective Date and Comment Period.
F. Relationship to Other Actions.

VI. Response to AAMA and CFAS Petitions.

VII. Granting of Petition to Consider Using 5th Percentile Female Dummy.

VIII. Rulemaking Analyses and Notices.

A. Executive Order 12866 and DOT Regulatory Policies and Procedures.
B. Regulatory Flexibility Act.
C. National Environmental Policy Act.
D. Executive Order 12612 (Federalism).
E. Civil Justice Reform.

IX. Request for Comments.

Appendix: Past Public Comments Related to Depowering Air Bags.

In 1984, the Department of Transportation issued a final rule requiring the installation of automatic protection (e.g., air bags, automatic belts, passive interiors) in passenger cars. 49 Fed. Reg. 28962; July 17, 1984. The Department took this step to increase the protection of vehicle occupants, especially unbelted ones. At the time, only 12.5 percent of occupants wore their safety belts, and only one state required all motorists to buckle up.

In 1991, Congress mandated the installation of air bags in both passenger cars and LTV's with a gross vehicle weight rating (GVWR) of 8,500 pounds or less. (LTV's generally include vans, pickup trucks, buses, and sport utility vehicles with a gross vehicle weight rating of 10,000 pounds or less). The Intermodal Surface Transportation Efficiency Act required that air bags be put in all new cars by the beginning of model year 1998 and in all new LTV's by the beginning of model year 1999.

Much has changed since 1984, and even since 1991. The cumulative production of air bag cars and LTV's reached the 10,000,000 mark for driver air bag vehicles during model year 1992 and for dual air bag vehicles during model year 1995. Air bags are now standard equipment on most passenger cars and LTV's. As of the end of model year 1996, approximately 56 million air bag vehicles have been produced for sale in the United States.⁽¹⁾ Safety belt use has reached approximately 68 percent.⁽²⁾ Forty-nine States and the District of Columbia require the use of safety belts, and all jurisdictions require the use of child safety seats. While males account for a sizable majority of the nonusers of safety belts, females still account for 40 percent of the nonusers.⁽³⁾

NHTSA estimates that air bags have deployed more than 800,000 times in crashes and have saved approximately 1,664 lives (164 passengers and 1,500 drivers) as of November 1996. Unfortunately, air bags also have fatally injured at least 32 children, 1 adult passenger, and 19 drivers in low severity crashes in the United States. Apart from the nine fatally-injured infants (included in the figure of 32 above), most of the fatally-injured occupants were unbelted. Thus, while the number of people being saved by air bags is growing annually, so is the much smaller, but significant number of people being fatally injured by air bags.

A. How Air Bags Work.

When a vehicle has a frontal impact, its occupants begin to move forward in response to pre-impact braking or the deceleration of the vehicle during the impact. If unrestrained, front-seat occupants will move forward in a fraction of a second and hit the steering wheel, dashboard or windshield. To move into place in time to catch the occupants in moderate and high speed crashes, air bags must inflate very quickly--faster than the blink of an eye.

To ensure that the air bag provides enough resistance to keep large as well as small occupants from "bottoming out" the air bag and hitting the vehicle interior, the amount of gaseous pressure within air bags must be carefully modulated. This is done by controlling both the rate at which gas is pumped into the air bag as well as the rate at which the gas is released from the air bag through vents or the porosity of the fabric.

An example from a non-automotive context will help to show the importance of modulating the air pressure in air bags. Vented air cushions are sometimes used by stunt performers who jump or dive from a great height to absorb the energy of their fall. If the vents don't allow enough of the pressure in the cushion to be released as the performer hits it, the cushion will be too rigid and will fail to absorb enough of the performer's energy, causing injury. On the other hand, if the vents release too much pressure, the cushion will "bottom out," thus allowing the performer to strike the ground, also causing injury.

B. Circumstances of Air Bag Fatalities.

Air bags need time, and space, to inflate. The sudden release of energy by an inflating air bag can harm some front seat occupants, particularly if they are too close to the air bag at the time of deployment. Properly restrained occupants of a vehicle seat moved back from the dashboard as far as possible, and even most unrestrained teenagers and adults, will meet the air bag after the initial, sudden release of energy. However, some occupants either start out very close to the steering wheel or dashboard or end up there. Most child fatalities attributed to an air bag fall into one of two groups: (1) infants riding in rear-facing infant seats, thus placing them very close to the air bag at the time of deployment, or (2) older children riding forward-facing without any type of restraint, thus allowing them to slide forward during pre-crash braking so that they were too close to the air bag when it deployed. A majority of the fatally-injured drivers were short-statured women who moved the driver's seat forward. More than half of the fatally-injured drivers were not using any type of restraint.

The number of air bag fatalities and the likelihood of those fatalities must be carefully compared to the likelihood of other related events in evaluating solutions to the causes of those fatalities.

A. Frontal Impacts. Frontal impacts are the number one fatality and injury-causing mode of crash, resulting in 64 percent of all driver and right-front passenger fatalities and 65 percent of all driver and right-front passenger AIS 2-5 injuries. (AIS 2-5 stands for Abbreviated Injury Scale levels of moderate to critical injuries.) The estimated fatality and injury totals for 1994 are shown below. The injuries are those for National Accident Sampling System-Crashworthiness Data System (NASS-CDS) towaway accidents only. (See table below.)

	Drivers	Right Front Passengers	Total
Fatalities	13,437	3,814	17,251
Injuries	124,484	30,299	154,783
Total	137,921(4)	34,113(5)	172,034

B. Air Bags: Lives Saved, and Lives Lost.

As the agency has confronted the problem of low speed fatalities and injuries from air bags, it has faced a serious dilemma. On the one hand, air bags have proven to be highly effective in reducing fatalities, and are resulting in substantial net benefits in terms of lives saved. The agency estimates that, to date, air bags have saved 1,664 drivers and passengers (1,500 drivers and 164 passengers).⁽⁶⁾ Current air bags could save an estimated slightly more than 3,000 lives each year in passenger cars and light trucks when all cars on the road are equipped with dual air bags.

At the same time, air bags are actually causing fatalities in some situations, especially to children. As of November 30, 1996, NHTSA's Special Crash Investigation program had identified 32 crashes in this country in which the deployment of the passenger air bag resulted in fatal injuries to a child. The agency has examined all air bag cases with child fatalities in its Fatal Accident Reporting System (FARS) and believes it has identified all cases involving air bag-related fatalities. One adult passenger has been fatally injured (a woman in her 90's). On the driver side, 19 drivers⁽⁷⁾ have been fatally injured in this country. (See table below.)

	Drivers	Right Front Passengers	Total
Lives Saved	1,500	164	1,664
Fatalities Caused	19	33	52
Net Lives Saved	1,481	131	1,612

Passenger Fatalities. The annual number of fatalities involving children is steadily growing; all have occurred in 1993 and later calendar years. As noted above, 32 children have been fatally injured to date. (See tables below.)

It appears that the children most at risk are infants in rear-facing infant restraints and children not using any type of restraint. All of the infant fatalities (9) involved infants in rear-facing child seats. Most of the other children were not using any type of safety restraint. Of those other children, 18 were unrestrained, two more were wearing only the lap belt with the shoulder belt behind them, and two were wearing a lap and shoulder belt at the time of the crash. In addition, there was a one-year-old child who was fatally injured while riding in a child seat that was not belted to the vehicle seat. (See table below.)

Most children were either infants or children aged 4-7 years old. (See table below.)

The crashes in which the children were fatally injured involved pre-impact braking, and occurred at relatively low speeds. Infants in rear-facing child seats are very close to the dashboard even before pre-impact braking. As to almost all of the older children, the nonuse, or improper use of safety belts in conjunction with pre-impact braking resulted in their forward movement such that they were very close to the instrument panel and the air bag system when the air bag deployed. Because of this proximity, the children appear to have sustained fatal head or neck injuries from the deploying passenger air bag.

In addition to the 32 children who have been fatally injured during passenger air bag deployments, as noted above, one adult, a woman in her 90's, sustained a fatal injury that appears to be due to an air bag deployment.

Driver Fatalities. As of November 15, 1996, NHTSA's Special Crash Investigation (SCI) program had identified 19 minor to moderate severity crashes in which fatal injuries to the driver were associated with the deployment of the driver air bag.⁽⁹⁾ The data suggest that unrestrained small-statured and/or older drivers are more at risk than other drivers from a driver air bag. (See tables below.) The agency notes that older drivers are more at risk than younger drivers under a wide range of crash circumstances, regardless of type of restraint used.

NHTSA notes that these driver fatalities are very rare in comparison to the number of vehicles equipped with driver air bags and to the number of drivers saved by air bags. Further, NHTSA notes that the last reported fatality in the United States of a female driver 5 feet 2 inches or shorter in an air bag deployment occurred in November 1995, 13 months ago.

Proper belt use is important. Ten of the 19 drivers were known to have been unrestrained at the time of the crash. Of the six persons properly using both lap and shoulder belts, two appeared to be out of position (slumped over the wheel due to medical conditions). (See tables below.)

Comparison of Passenger and Driver Air Bag Fatalities. Several comparisons need to be drawn between the trends and patterns of child fatalities and the apparent trends and patterns of driver fatalities. The annual number of child fatalities is clearly growing steadily as the number of deployments increases. The annual number of adult fatalities does not appear to be growing. If anything, it appears to be decreasing, based on currently identified fatalities. (See tables below.)

Most child fatalities (24 of 32) have occurred in model year 1994 and 1995 vehicles. In contrast, only 4 of the 19 driver fatalities have occurred in a vehicle manufactured after model year 1992. The absence of fatalities in recent model year vehicles appears even more pronounced in the case of women 5 feet 2 inches or shorter. Only one woman 5 feet 2 inches or shorter has died in a post model year 1992 vehicle.⁽¹⁰⁾ Most fatalities of short-statured women occurred in model year 1990-1992 vehicles. (See tables below.)

Potential Number of Persons Saved versus the Potential Number Fatally Injured by Current Air Bags. The dilemma faced by NHTSA, and ultimately the public, is how to address the problem of low speed fatalities from air bags while preserving their substantial life-saving benefits. Based on analyses of real world data, NHTSA estimates that if all passenger cars and light trucks on the road today had current air bags, there would be more than 3,000 lives saved each year, as compared to a no-air-bag fleet (assuming current belt use rates). More than two-thirds of the persons saved would be persons not using any type of safety belt.

On the driver side, 616 belted drivers and 1,686 unbelted drivers would be saved, for a total of 2,302 lives saved. This is a net figure, i.e., it accounts for the possibility of 25 drivers being fatally injured annually by an air bag. Given that the average annual rate of driver fatalities for the last five years appears to be three, and that the annual rate does not appear to be increasing, the projected figure of 25 may be somewhat overstated.

The potential number of lives saved by passenger air bags is much smaller than driver air bags primarily because the passenger seat is occupied much less frequently than the driver seat.If all passenger cars and light trucks had current passenger air bags, the agency estimates that 223 belted and 491 unbelted passengers aged 13 and above would be saved annually, for a total of 714 lives.

However, this figure of 714 would be partially offset by air bag-related fatalities involving children 12 and under. If current rates of child fatalities were experienced in an all-air-bag fleet, 128 children would be fatally injured by air bags annually, again assuming no technological improvements, changes to air bags, or behavioral changes by vehicle operators (e.g., ensuring that any children placed in the front seat properly use occupant restraints or, preferably, placing children in the rear seat). The figure of 128 includes 90 forward-facing children, most of whom would be unbelted, and 38 infants in rear-facing child restraints.

NHTSA emphasizes that this and the other rulemaking proceedings and related efforts are intended to ensure that risks of adverse effects of air bags are reduced so that these theoretically projected air bag fatalities do not materialize, while the potential benefits of air bags are retained, to the maximum extent possible. Thus, the agency anticipates, e.g., that these other actions will result in proper use of restraints by increased numbers of people and that the number of children fatally injured would not be so high as 128. However, the agency does not have a basis for determining the exact effect.

A. The Early Years. The potential that air bags could cause fatalities and the means of reducing that potential have been discussed for as long as air bags have been considered as an important method for reducing fatal and serious injuries in frontal crashes. Vehicle manufacturers raised the possibility of such fatalities at the first agency public meeting on the subject of automatic restraints, held in 1969.

From the very beginning, the industry, the agency and the research community have identified and explored possible solutions. For example, in the early and mid-1970's, various vehicle manufacturers reported favorable results in testing the ability of various dual level or variable inflation systems for air bags to address the problem of out-of-position children. In 1980, NHTSA informed the industry about its analysis of a number of possible technological solutions, including dual-inflation air bags, chambering air bags and top-mounted air bags. The July 11, 1984 Final Regulatory Impact Analysis (FRIA) for the 1984 final rule requiring the installation of automatic occupant restraints in passenger cars (49 Fed. Reg. 28962; July 17, 1984) identified a variety of possible solutions:

The agency has analyzed the effect of air bag systems on various ages and sizes of occupants, with a particular emphasis on the small child.

* * *

[A] small number of children could in fact be at greater risk from the air bag induced trauma than . . . from the effects of the crash itself.

The automobile industry, the research community, and NHTSA have done a tremendous amount of work over the years in trying to assess the air bag's potential for injury to out-of-position occupants, and to assess the probability of those injuries occurring in the real world.

At this time, air bag technology could be likened to a drug with great potential lifesaving and injury reducing capability, but with some limited adverse side effects for some (out-of-position children). In the past few years child restraint legislation has been enacted in nearly all of the states. This has the effect of reducing the probability that a child would be out-of-position to levels below that used in previous studies. Nonetheless, any air bag design should attempt to minimize the probability of a child being injured, regardless of position, while maintaining the large potential lifesaving benefits for children and other occupants.

In summary, the agency concludes that although air bags, on isolated occasions, may cause injuries that may not have otherwise occurred, their overall safety benefits far outweigh this chance occurrence. Air bags are no different from other safety devices in this regard.

(FRIA, pp. III-8 to 10)

The FRIA then listed a variety of potential technological means for addressing the problem of injuries associated with air bag deployments:

• a dual level inflation system whose operation is based on impact speed (a low level for lower impact speeds and a higher one for higher speeds);

• a dual level inflation system whose operation is based on a switch in the vehicle seat or elsewhere that measures occupant size or weight and senses whether an occupant is out of position (a low level for out-of-position occupants and a higher one for properly positioned occupants);

• A dual level inflation system whose operation is based on an electronic proximity detector in the dashboard (a low level if the occupant is near the dashboard and a higher one if the occupant is farther away); and

• other technological measures such as bag shape and size, instrument panel contour, aspiration, and inflation technique.

B. The Last Five Years. Over the last five years, NHTSA has taken a variety of steps to alert the public to the dangers posed by air bags to children and to explore measures for reducing and even eliminating those dangers. The steps taken in 1991-1995 were recounted in an NPRM published by the agency on August 6, 1996. 61 Fed. Reg. 40784.

In the August 1996 NPRM, the agency proposed several amendments to Standard No. 208, Occupant Crash Protection, and Standard No. 213, Child Restraint Systems, to reduce the adverse effects of air bags, especially those on children. The agency explained that eventually, either through market forces or government regulation, it expects "smart" passenger air bags to be installed in passenger cars and light trucks to mitigate these adverse effects. NHTSA indicated that, for purposes of the NPRM, it considered smart passenger air bags to include any system that automatically prevents an air bag from injuring the two groups of children that experience has shown to be at special risk from air bags: infants in rear-facing child seats, and children who are out-of-position (because they are unbelted or improperly belted) when the air bag deploys.

NHTSA proposed that vehicles lacking smart passenger air bags would be required to have new, attention-getting warning labels and permitted to have a manual cutoff switch for the passenger air bag. By limiting the labeling requirement to vehicles without smart passenger air bags, NHTSA hoped to encourage the introduction of the next generation of air bags as soon as possible. NHTSA proposed to define smart air bags broadly to give manufacturers flexibility in making design choices. The agency requested comments concerning whether it should require installation of smart air bags and, if so, on what date such a requirement should become effective. NHTSA also requested comments on whether it should, as an alternative, set a time limit on the provision permitting manual cutoff switches for passenger air bags in order to assure the timely introduction of smart passenger air bags. Finally, the agency proposed to require rear-facing child seats to bear new, enhanced warning labels. In a section in the August 1996 NPRM titled "Future Agency Considerations," the agency also provided a discussion of possible technological changes to address the forcefulness of air bag deployment, ongoing agency efforts to evaluate the effects of such changes, and possible future agency regulatory actions.

C. Recent Petitions for Rulemaking

Two weeks before the agency published its NPRM, the Parents' Coalition for Air Bag Warnings submitted a petition requesting the agency to commence a rulemaking proceeding to require that the following warning label be placed on dashboard of vehicles with passenger air bags: "WARNING: DO NOT SEAT CHILDREN IN THE FRONT PASSENGER SEAT. AIR BAG DEPLOYMENT CAN CAUSE SERIOUS INJURY OR DEATH TO CHILDREN."

After the agency's publication of the August 1996 NPRM, the American Automobile Manufacturers Association (AAMA) submitted a petition for rulemaking requesting that NHTSA immediately announce, by means of a "direct final rule," an amendment to Standard No. 208 to replace the current 30 mph unrestrained dummy barrier crash test requirement with a sled test protocol incorporating a 143 millisecond standardized crash pulse. The petitioner contended that the standard's current requirement "directly dictates the level of the air bag's inflator power and it is the level of inflator power that unnecessarily increases the risk of injury to vehicle occupants during air bag deployment." AAMA also requested that the agency separately issue a notice of proposed rulemaking to propose requirements to improve the safety of drivers and passengers who are extremely close to the air bag at the time of deployment, based on the latest International Standards Organization (ISO) test practices. AAMA recommended the use of the Hybrid III small female dummy in the driver position and appropriate child dummy in the passenger position.

On September 1, 1996, Ms. Anita Glass Lindsey petitioned the agency to commence rulemaking to specify the use of a test dummy representing a 5th percentile female⁽¹²⁾ in testing the performance of safety belts and air bags. Currently, Standard No. 208 specifies the use of only a 50th percentile male test dummy.⁽¹³⁾

On September 17, 1996, the National Transportation Safety Board (NTSB) issued a number of safety recommendations to NHTSA for reducing the problem of child fatalities caused by air bags. These recommendations are as follows:

1. Immediately evaluate passenger air bags based on all available sources, including NHTSA's recent crash testing, and then publicize the findings and modify performance and testing requirements, as appropriate, based on the findings of the evaluation.

2. Immediately revise Federal Motor Vehicle Safety Standard 208, Occupant Crash Protection, to establish performance requirements for passenger air bags based on testing procedures that reflect actual accident environments, including pre-impact braking, out-of-position child occupants (belted and unbelted), properly positioned belted child occupants, and with the seat track in the forward-most position.

3. Evaluate the effect of higher deployment thresholds for passenger air bags in combination with the recommended changes in air bag performance certification testing, and then modify the deployment thresholds based on the findings of the evaluation.

4. Establish a timetable to implement intelligent air bag technology that will moderate or prevent the air bag from deployment if full deployment would pose an injury hazard to a belted or unbelted occupant in the right front seating position, such as a child who is seated too close to the instrument panel, a child who moves forward because of pre-impact braking, or a child who is restrained in a rear-facing child restraint system.

5. Determine the feasibility of applying technical solutions to vehicles not covered by NHTSA's proposed rulemaking of August 1, 1996, to prevent air bag-induced injuries to children in the passenger position.

On November 8, 1996, the Center for Auto Safety (CFAS) petitioned the agency to amend Standard No. 208 to specify that a vehicle's air bags must not deploy in a crash if the vehicle's change of velocity is less than 12 mph. CFAS noted that many of the crashes resulting in air bag fatalities, especially those of children, involved very low changes in vehicle velocity. CFAS also petitioned the agency to institute investigations of several vehicle models for alleged defects related to air bag deployment.

On November 13, 1996, the AAMA submitted a letter that modified the proposal in its August 1996 petition for rulemaking. In place of the 143 millisecond standardized crash pulse, AAMA requested a sled test protocol incorporating a 125 millisecond standardized crash pulse.

Finally, on November 20, 1996, CFAS and Public Citizen petitioned the agency to begin rulemaking to require dual inflation air bags. These bags would inflate more slowly, and thus less aggressively, than current air bags in low-speed crashes. In higher-speed crashes, they would inflate at the same rate as current air bags. The petitioners assert that their proposal is the best solution in the near future and is superior to depowering, since depowering involves "some trade-off in safety protection and will not add significant protection for unrestrained children."

NHTSA is implementing a comprehensive plan of rulemaking and other actions (e.g., primary enforcement of State safety belt use laws) addressing the adverse effects of air bags. As part of that plan, NHTSA is issuing three separate, but related, notices today. Each notice is intended to ensure that some or all or the risks are reduced, and benefits retained, to the maximum extent possible. They provide immediate and/or interim solutions to the problem. A later notice, a proposal to require smart air bags, would provide a permanent solution.

In this notice, NHTSA is proposing to temporarily amend the agency's occupant crash protection standard to help reduce the fatalities and injuries that current air bags are causing in relatively low speed crashes to small, but growing numbers of children, and occasionally to adults. Based on agency research and analysis regarding the optimal range of air bag depowering, the agency has tentatively concluded that an average depowering of 20 to 35 percent would reduce the risk of fatalities in low speed crashes, while substantially preserving the life-saving capabilities of air bags in higher speed crashes.

The agency is considering the adoption of either, or both, of two different approaches that would permit or facilitate an approximate 20 to 35 percent average depowering of current air bags. One approach would be to temporarily reduce the stringency of the chest acceleration requirement that an unbelted dummy must meet in a crash test at speeds up to 30 mph. The other approach would be to temporarily adopt the AAMA's modified proposal for a sled test protocol incorporating a 125 millisecond standardized crash pulse.

NHTSA is seeking comments and information concerning the relative desirability of these two approaches, including supporting data from industry with respect to the sled test. It is also requesting comments on the appropriate duration of such a temporary amendment. NHTSA anticipates that it would remain in effect for both the passenger and driver seating positions until smart air bags are installed pursuant to a mandated phase-in schedule, which will be the subject of a separate rulemaking proceeding. Finally, comments are sought on whether the same or different requirements should apply to the passenger and driver positions.

The other rulemaking actions addressing the adverse side effects of air bags are as follows:

• Based on the August 1996 NPRM, the agency issued on November 22, 1996, a final rule amending Standards No. 208 and No. 213 to require improved labeling on new vehicles and child restraints to better ensure that drivers and other occupants are aware of the dangers posed by passenger air bags to children. The labeling places particular emphasis on placing rear-facing infant restraints in the rear seats of vehicles with operational passenger air bags. 61 Fed. Reg. 60206; November 27, 1996. The new labels are required on vehicles not equipped with smart passenger air bags beginning February 25, 1997, and on child restraints beginning May 27, 1997.

• Based on the same NPRM, the agency is issuing a final rule extending until September 1, 2000, a provision in Standard No. 208 permitting vehicle manufacturers to offer manual cutoff switches for the passenger air bag for new vehicles without rear seats or with rear seats that are too small to accommodate rear-facing infant restraints.

• The agency also is issuing an NPRM proposing to permit motor vehicle dealers and repair businesses to deactivate, upon the request of consumers, driver and passenger air bags that do not meet the agency's criteria for smart air bags. Final action is expected in early 1997.

• In addition to these actions, NHTSA will issue a separate supplemental NPRM (SNPRM) to require a phasing-in of smart air bags, beginning on September 1, 1998, and to establish performance requirements for those air bags. The proposal will be issued in early 1997.

The next two tables summarize the rulemaking actions included in the agency's comprehensive program to address these air bag problems:

In addition to these actions, the agency is participating with automobile manufacturers, air bag suppliers, insurance companies and safety organizations in a coalition effort to address the adverse effects of air bags by increasing the use of safety belts and child seats. Substantial benefits could be obtained from achieving higher safety belt use rates. If the safety belt use rate were 75 percent in potentially fatal crashes instead of the current level of 52.6 percent, an additional 4,000 lives would be saved annually.

The coalition has a three-point program that seeks to educate the public about safety belt and child seat use, work with state and local officials to improve enforcement of safety belt and child seat use laws and seek the enactment of "primary" safety belt use laws. In States with "secondary" safety belt use laws, law enforcement officials are hampered in their ability to enforce the requirement to use safety belts because their inability to stop and ticket motorists for the sole reason of the motorists' failure to use their safety belts. A motorist may be ticketed by an official for such failure only if the official has a separate basis for stopping the motorist, such as the violation of a separate traffic law.

A 1995 NHTSA analysis of FARS data on restraint use among fatally injured motor vehicle occupants from 1983 to 1994 indicates that primary enforcement is the most important aspect of a safety belt use law affecting the rate of safety belt use. For virtually all states with a primary enforcement law, statistically significant increases associated with the presence of such a law were detected using several different methods. The analysis suggests that the increase in use rates attributable to the enactment of a use law can be estimated to be (on the average) at least 25 percentage points, while the additional increase attributable to primary enforcement of the law is at least 15 additional percentage points. These increases in safety belt use translate into an estimated 12.6 percent decrease in fatalities in a state that enacts a safety belt use law, and an additional 5.9 percent decline in fatalities in a state that authorizes primary enforcement of the law.

State data support these findings. On average, states with a primary safety belt law have usage rates that are 10-15 percentage points higher than states with secondary laws. In California and Louisiana, states which recently upgraded their laws to allow for primary enforcement, safety belt usage increased by 13 and 17 percentage points, respectively.

V. Depowering Air Bags

A. Results of NHTSA Test Program. To determine whether current air bags can be depowered to a degree that makes a significant contribution to reducing the risk of serious or fatal injury to occupants, especially children, without substantial loss of protection for teenagers andadults, the agency initiated the research testing and analysis program discussed in the August 1996 NPRM. NHTSA explained:

The agency has initiated a research testing and analysis program ... at the Vehicle Research and Test Center, the agency's in-house laboratory in Ohio. The program's objectives are to:

• Assess the performance of air bag systems in current production vehicles in particular crash conditions, including the effects on out-of-position children.

• Assess the level of improvement possible in out-of-position performance from changes to existing air bag components, including downloaded air bags, as well as newly developed pre-production systems.

• Provide visibility for air bag-related technology, thus promoting the rapid adoption of newer technologies that will help solve the out-of-position occupant injury problem.

The immediate focus of the program is on the passenger out-of-position problem as related to children. Several vehicle models have been selected based upon field accident investigations and air bag design characteristics. Both domestic and foreign vehicles are included in the selection. The test conditions include four different child positions similar to those recommended by ISO [International Standards Organization], and represent worst case occurrences. These tests will provide "baseline" performance of air bag systems when a child is an out-of-position occupant.

NHTSA is inviting vehicle manufacturers and air bag and component suppliers to provide state-of-the-art air bag systems. Systems that show significant improvements over baseline performance for out-of-position children will also be tested with adult-sized dummies in full-scale crash conditions required in Federal standards.

The test program will also address other aspects of air bag safety following the out-of-position child study. These include out-of-position driver tests, vehicle crash sensor testing, and testing of advanced air bag systems. The out-of-position driver testing will focus on small-sized female occupants who are sometimes injured due to the close proximity to the steering-wheel air bag system. Testing will continue into fiscal year 1997.

(61 Fed. Reg. 40784, at 40799; August 6, 1996.)

NHTSA has now tested the depowered air bags solicited from the vehicle manufacturers. The air bags had been depowered through the removal of certain amounts of propellant. While some of the air bags were depowered up to 60 percent, most of them were depowered an average of approximately 20 to 35 percent. However, their design (e.g., folding patterns and venting) had not been optimized for the reduced levels of power. As noted below, the agency believes optimization of the tested air bags would have significantly enhanced their performance.

NHTSA tested baseline air bags (i.e., air bags of current design) and depowered air bags on the passenger side in three different vehicles, and on the driver side in one vehicle.⁽¹⁴⁾ NHTSA conducted these tests using modified versions of recommended test procedures formally adopted and issued in early 1996 by the ISO for evaluating child restraint system interactions (ISO TR 14645) and out-of-position vehicle occupant interactions (ISO TR 10982) with deploying air bags. For the passenger air bags, the agency conducted various tests using out-of-position three-year-old and six-year-old child dummies and normally-positioned, belted and unbelted 50th percentile male dummies.⁽¹⁵⁾ For the driver air bags, the agency conducted various tests using out-of-position 5th percentile female dummies and normally-positioned, belted and unbelted 50th percentile male dummies. The agency also used computer-assisted mathematical modeling in an attempt to assess the effects of depowering on the forces experienced by occupants in air bag deployments.

The results of the agency's analysis of this testing, as well as other available information, are included in the Preliminary Regulation Evaluation (PRE)for this rulemaking. Portions of the PRE are summarized below.

B. Effects of Depowering and Optimizing.

Overview. The agency's testing and other available information⁽¹⁶⁾ indicated that depowering by an average of 20 to 35 percent substantially reduced injury measures for persons close to the air bag, especially out-of-position children, while producing only small increases in injury measures for adult dummies. In the agency's testing, depowering more than 35 percent resulted in more substantial increases in adult dummy injury measures with a large additional reduction in out-of-position child dummy injury measures for only the more aggressive air bags. Thus, it appears that depowering at levels more than an average of 35 percent could result in losing a significant portion of the benefits being provided by air bags without a commensurate reduction in child injury risk. (However, it is possible that some of today's air bags are so aggressive that they could, if optimized, be depowered by more than 35 percent without substantial losses in adult benefits.)

The reductions in injury measures achieved by depowering an average of 20-35 percent would contribute significantly to solving the problem created by overly aggressive air bags.⁽¹⁷⁾ While this average level of depowering would not eliminate all of the risk of serious injury to all persons currently at risk, it would eliminate much of the risk. The agency's other rulemaking actions would reduce the residual risk.

As noted above, the tested air bags were depowered, but not optimized. Had they been optimized, the injury measures for belted passengers would likely have decreased even more and those for belted drivers would likely have improved. Thus, they would have offered increased safety for belted occupants.⁽¹⁸⁾

Summary of Effects of Depowering on Air Bag-Related Fatalities for Particular At-Risk Occupant Groups

The ability of depowering to prevent air bag fatalities to occupants would vary depending on a number of factors, especially the location and belt use of the occupant. As shown in testing by the agency of passenger air bags, the forces exerted by a deploying air bag generally decrease as a function of increasing distance from the air bag module. Although the surface of an expanding air bag in its initial moments of inflation is potentially lethal, it rapidly changes within inches into an injury-preventing and life-saving surface as it inflates and moves away from its storage location. Thus, the farther away an occupant is from an air bag as it starts to inflate, the better off that occupant will be. While this is true for depowered as well as current air bags, depowering can significantly reduce the size of the zone within which serious injury is possible or likely.

Passengers. The at-risk groups are infants and young children. Properly belted, forward-facing children who are on a vehicle seat moved all the way back, should be at essentially no risk from a deploying, depowered air bag, even if they are leaning forward while belted. Moderately out-of position, forward-facing children would receive substantial benefits. Severely out-of position, completely unbelted forward-facing children would receive some benefits. Given their proximity to the air bag, infants in rear-facing child restraints would likely receive only small, unquantifiable benefits from depowered air bags.⁽¹⁹⁾

Drivers. To the extent that there is an at-risk group, it is short-statured women. Short, belted drivers on a vehicle seat moved as far back as their stature permits would receive substantial benefits, particularly with respect to neck injuries. They are not likely to move as far forward as unbelted drivers during pre-crash braking and during the initial stages of a crash. Benefits for unbelted drivers on a vehicle seat moved all the way forward would depend on the drivers' proximity to the air bag at the time of deployment. If they are at least two or three inches away at the time of deployment, they should receive some benefits from depowering with respect to chest and head injuries. Depowering should help all drivers with respect to arm injuries.

Overall Effects of Depowering. The PRE estimates the potential overall effects of depowering on all forward-facing children, teenage and adult occupants under the two alternative proposals, the 80 g alternative and the generic sled test alternative. Both proposals would produce a mixture of benefits and disbenefits, with the benefits primarily accruing to children and belted teenage and adult occupants, and the disbenefits primarily accruing to unbelted teenage and adult occupants.

The magnitude of the benefits and disbenefits are estimated in the PRE by two different methods. Method One includes only fatalities, while Method Two includes fatalities and serious injuries. The results of Method One, which produces slightly smaller upper end values for lives saved and for foregone savings of lives, are discussed below.

1. Passenger Air Bags.

Child Passengers. Older, Forward-Facing Children. Depowering could prevent a significant number of the 90 annual fatalities projected above for forward-facing children⁽²⁰⁾ in an all air bag fleet for passenger cars and LTV's. The PRE estimates that 39 of the projected 90 fatalities could be prevented by depowering air bags by an average of 20 to 35 percent. This includes all of the lap and shoulder belted children who might otherwise be fatally injured and most of the moderately out-of-position children.⁽²¹⁾ With the additional depowering possible under the generic sled alternative,⁽²²⁾ up to 83 of the projected 90 fatalities could be prevented since more of the severely out-of-position children could be benefited. Thus, depowering would make it safe, from the standpoint of the air bag, to place a child in the front seat when necessary, assuming that the child was properly restrained in a vehicle seat that was moved all the way back. The agency emphasizes that, even in the absence of an air bag, the rear seat is a significantly safer place for children to ride than the front seat.

Rear-Facing Children (Infants). Based on HIC reductions achieved in testing the effects of depowered air bags on three-and six-year-old dummies, the agency believes that depowering could prevent the death of some of the 38 projected fatalities of infants. However, for reasons explained below, the agency cannot quantify those savings.

As noted above, the agency did not perform any testing of depowered air bags with infants in rear-facing infant seats. Thus, the agency does not have any baseline versus depowered air bag data for rear-facing child restraints to estimate the potential benefits of depowering. However, HIC data from the testing of severely out-of-position three- and six-year-old children indicate that HIC was substantially reduced by depowering, but not typically below the assumed infant injury reference value of 500 HIC. HIC data are relevant because the primary cause of rear-facing infant fatalities in air bag deployments has been skull fractures. Since it is not possible at this time to make appropriate adjustments to reflect greater susceptibility of infants to fatal head injury, the HIC data for dummies representing older children could not be used to estimate potential benefits of depowering for infants. The agency has not made a specific, quantified estimate because of its roughness and therefore its questionable value.

Teenage and Adult Passengers. Depowering air bags to an average of 20 to 35 percent would likely benefit belted teenage and adult passengers on balance, but could necessitate foregoing the opportunity to save some unbelted teenage and adult passengers.⁽²³⁾ These estimates are based on chest g measures because, as noted in the PRE, chest g's are the most important measure for assessing the effects on teenagers and adults, since chest g's appear to have a stronger relationship to fatality risk than HIC. Further, the HIC increases due to depowering in this range were not that significant.

Belted Teenage and Adult Passengers. The agency's PRE assumes a 2.4 g decrease in chest g's for belted passengers under the 80 g alternative, using an air bag that had been depowered but not optimized. This assumption was based on test results showing a 2.4 g decrease in chest g's, although mathematical modeling predicted almost no change for belted passengers. Under the generic sled test alternative, a decrease of 1.9 chest g's is assumed, based on mathematical modeling. Both decreases would result in saving additional lives compared to current air bag designs.

As noted above, NHTSA believes that a greater decrease in chest g's, and therefore a greater increase in life-saving potential, would have occurred had the air bags not only been depowered, but also optimized for the new power level. The depowered air bags tested by NHTSA were not optimized in ways that would likely have reduced the chest g's even more. For example, the air bags were not optimized with respect to their venting rates.

The agency believes that it is unlikely that the vehicle manufacturers would depower their air bags without also optimizing them. NHTSA believes that the manufacturers would, out of reasonable prudence, do both.

This is significant because real world data from Australia regarding the performance of depowered driver air bags optimized for belted occupants suggests that depowering and optimizing current U.S. air bags could significantly increase the effectiveness of air bags for belted occupants and lead to large savings of lives. Those data, drawn from crashes involving Holden passenger cars,⁽²⁴⁾ indicate that air bags with lap/shoulder belts reduced AIS 2+ injuries to drivers by 39 percent compared to lap/shoulder belts alone. By comparison, current U.S. air bags have an AIS 2+ effectiveness of 22 percent when lap and shoulder belts are worn. According to the PRE:

The air bag systems in the Commodore are designed to deploy as unaggressively as possible while still providing the necessary protection to occupants of different size, weight and sex who will be potentially involved in a variety of collisions. Great efforts have been taken in the development of the inflators and cushions to ensure they present as little risk as possible to occupants during inflation. Since the air bags have been designed to operate in conjunction with the safety belts, they are only required to decelerate the occupant's head and upper torso, as the primary load path is through the belts. This is fundamentally different from many other air bag designs, especially those used to protect unrestrained occupants. Systems optimized to protect unrestrained occupants typically utilize high-performance inflators in conjunction with cushions with low venting rates. This combination ensures that the air bags are sufficiently stiff to decelerate unbelted occupants.

(Page V-1)

If such increased effectiveness could be obtained for belted passengers, it would offset a significant portion of the potential adverse impact of depowering estimated below on unbelted passengers. As discussed in the PRE, current NHTSA analyses indicate that air bags in this country are 8.5 percent effective in reducing belted fatalities. If the relationship in overall effectiveness of the Holden bag to the U.S. air bags for AIS 2+ injuries were the same for fatalities, the effectiveness of U.S. air bags for preventing fatalities to belted occupants could be as high as 15 percent. If depowering and optimizing U.S. air bags increased their effectiveness to that level, large savings in the lives of belted occupants could result.

The agency seeks comments, on a model-by-model basis, if possible, from the vehicle manufacturers on what specific optimization measures they would adopt and on whether such optimization could be accomplished and incorporated in production air bags within the time frame projected by the vehicle industry for introduction of the depowered air bags. As noted below, AAMA projected that its members could begin introducing depowered air bags within 6-9 months and complete the process across their fleets within a year after those first introductions. NHTSA solicits comments as to what effect, if any, efforts to optimize these air bags prior to their introduction might have on the schedule for their introduction. Comment is also sought whether adoption of the sled test suggested by AAMA would enable vehicle manufacturers to accelerate the introduction of optimized and depowered air bags. The agency also requests comments on what effects, if any, the optimization of air bag performance for the benefit of belted occupants would have on air bag effectiveness for unbelted occupants. Finally, comment is sought on the Holden data and the reasonableness of the assumption in the PRE that effectiveness of U.S. air bags in reducing belted fatalities could be raised substantially in the next several years through depowering and optimizing.

Unbelted Teenage and Adult Passengers. Depowering could necessitate foregoing the opportunity to save a significant number of unbelted teenagers and adults. The PRE estimates that, as a result of a significant increase in chest g's associated with depowering by an average of 20 to 35 percent under the 80 g alternative, there could be a reduction of between 86 and 280 unbelted passengers who would have otherwise been saved by current air bags. This reduction reflects an assumed average increase of 11 g's in the chest g's for unbelted passengers as a result of depowering, but not optimizing air bags. This assumption was based on limited test results showing an 11 g increase in chest g's at 30 mph. Mathematical modeling predicted a slightly lower increase. With greater depowering under the generic sled test alternative, it was assumed that chest g's would increase by 22 g's, based on sled tests and mathematical modeling. That increase would result in a potential loss of savings of 115 to 336 unbelted passengers.

It should be noted, however, that AAMA does not anticipate such losses. AAMA provided an estimate of the effects of depowering, based on NASS data, a number of analytic assumptions, and sled/barrier test results. That organization estimates the potential savings of 30 to 200 small adults per year due to increased effectiveness of passenger and driver air bags for those persons and the potential loss of up to eight large adults annually. The agency seeks comment from AAMA on how it calculated those figures.

Further, to the extent that increased numbers of people use their safety belts, the potential losses in savings of unbelted passengers would not materialize. While increasing safety belt use would reduce the benefits of depowering, by reducing the size of some groups (i.e., unbelted children and drivers) vulnerable to air bag fatalities, there would be very large increases in the number of people saved by occupant restraints of one type or another. As noted above, if the safety belt use rate were 75 percent in potentially fatal crashes instead of the current level of 52.6 percent, an additional 4,000 lives would be saved annually. NHTSA plans to work vigorously with the States to increase safety belt use through public education and authorizing primary enforcement of safety belt use laws.

Safety Tradeoffs. NHTSA has carefully considered the potential tradeoffs implicit in depowering passenger air bags. Given the wide range of the above estimates concerning unbelted passengers, the agency believes that the net effect of depowering on safety could be positive. However, even if the net effect were negative, the agency believes that the opportunity to save a significant number of children who would otherwise be fatally injured by air bags justifies foregoing the opportunity to save some unbelted passengers. There are several reasons for this policy choice.

First, it is not acceptable that a safety device cause a significant number of fatalities in circumstances in which fatal or serious injuries would not otherwise occur. In making this statement, the agency draws a distinction between air bags which are fatally injuring young children in low speed crashes in which the other vehicle occupants are uninjured, and other safety devices which may on occasion unavoidably substitute one type of injury for another type that would occur in their absence (safety belts are a good example).⁽²⁵⁾ Those fatalities are particularly unacceptable in light of the agency's analysis showing that depowering air bags can significantly reduce the number of children being fatally injured by air bags.

Second, it is also particularly unacceptable that the vehicle occupants being fatally injured are young children, and that the number of those deaths is steadily growing. In confronting the possibility of inevitable short-term safety tradeoffs between young children and unbelted occupants over 12 years of age, the agency believes that greater weight must be placed on protecting young children. NHTSA has always given a high priority to protecting children. The agency's activities related to school bus safety standards are an example of this policy.

A major reason for giving priority to protecting young children is that they are less mature than teenagers and adults and thus less able to exercise independent judgment, assess the risks and take action to improve their safety. The young children are more dependent on the judgment and actions of other persons. The oldest of the 32 children who have been fatally injured by an air bag was nine years old, and most of the children have been much younger. Nineteen were four to seven years old and nine were infants. Conversely, the unbelted teenagers and adults who might not be saved as a result of depowering can take action on their own to protect themselves by simply buckling their safety belts as required by the laws of 49 States and the District of Columbia.

Notwithstanding the justifications for making the safety tradeoffs, NHTSA is concerned about them. It is because of the possibility of disbenefits, especially for unbelted occupants, that the agency is proposing to make only a temporary change in Standard No. 208 to permit or facilitate the depowering of air bags. The agency will shortly issue a proposal to require a phase-in of smart air bags. Requiring smart air bags would not only enable the agency to make depowering a temporary measure, but would also ensure that the problem of adverse effects from air bags is fully addressed, and that air bags achieve their full safety potential for protecting a wide variety of vehicle occupants over an appropriate range of vehicle speeds.

2. Driver Air Bags. Analysis of the net effect of depowering driver air bags is more difficult and therefore less precise largely because the agency has conducted fewer tests of depowered driver air bags and because the test results for the unbelted drivers are a mixture of small increases and decreases in chest g's. Nevertheless, the agency believes that depowering driver air bags would enhance safety. As noted above, belted short drivers who move their seat as far back as their stature permits, would benefit substantially from depowering. Belted drivers, in general, should benefit as well since depowering appears to allow a better "tuning" of the combined safety belt-air bag system for belted occupants. Unbelted, out-of-position short drivers could receive some benefit as well. As a result, there would be some reduction in the projected figure of 25 driver fatalities per year.

Belted Drivers. Depowering alone increased the chest g's for belted drivers in NHTSA's vehicle testing. Although the tests showed a 7 g increase at 35 mpg, there appears to be no logical reason for such an increase. In the same test, chest g's decreased for the belted passenger dummy. Further, modeling suggested only a marginal increase of 2 g. The PRE assumes a 2 g increase for belted drivers under the 80 g alternative. Under the generic sled test alternative, chest g's go up or down at different speeds with the net result that there would be no change in overall fatalities for depowered, but not optimized, air bags.

As in the case of passenger air bags and belted passengers, the agency believes that the data concerning the air bags in the Australian Holden passenger car show that optimizing as well as depowering air driver bags would produce a more favorable result for belted drivers than the depowered air bags tested by NHTSA. Since most of the Holden data related to driver air bags instead of passenger air bags, the agency has good reason to be even more confident about the implications of the Holden data for belted drivers in this country. With optimization, the agency believes that, instead of an increase in chest g's under the 80 g alternative or no change under the generic sled test alternative, a decrease is likely. If depowering and optimizing U.S. driver air bags increased their effectiveness to as much as 15 percent, the savings would be 471 drivers.

Unbelted Drivers. Depowering by an average of 20 to 35 percent under the 80 g alternative appears to slightly increase the chest g's of unbelted drivers. It is believed that the energy absorbing steering column is the reason that chest g's do not increase in proportion to the amount of depowering. In vehicle tests with depowered air bags, chest g's increased by 2 g at 30 mph, but decreased by almost 3 g's at 35 mph. The results of modeling were mixed also, but consistent with the vehicle test results. Modeling predicted a slight increase at 30 mph and decrease at 35 mph. Since there was an increase at some speeds, the PRE assumes a 2 g increase under the 80 g alternative. Based on that increase, the PRE estimates a possible loss in savings of 9 to 41 unbelted drivers. Under the generic sled test alternative, the PRE assumed a 10 g increase based on modeling. That increase suggests a resulting loss of 221 to 650 unbelted drivers.

As noted above, there is reason to believe that these losses might not occur. AAMA estimates the potential savings of 30 to 200 small adults per year due to increased effectiveness of passenger and driver air bags for those persons and the potential loss of up to eight large adults annually. Further, to the extent that increased numbers of people use their safety belts, the potential losses in savings of unbelted passengers would not materialize. NHTSA plans to work vigorously with the States to increase safety belt use through public education and authorizing primary enforcement of safety belt use laws.

Arm Injuries. The agency believes that depowering would lead to a significant reduction in driver arm injuries associated with air bag deployments. Compared to MY 1994 vehicles, depowering air bags by an average of 20 to 30 percent could reduce AIS 2-3 arm injuries from 25,006 to 16,254, a reduction of about 8,800 injuries.

Safety Tradeoffs. NHTSA has carefully considered the potential tradeoffs implicit in depowering driver air bags. Despite the wide range of the above estimates concerning unbelted drivers, the agency believes that the net safety effect of depowering passenger air bags could be positive instead of negative. Even if the net effect were negative, the agency believes that the opportunity to avoid causing fatal injuries to some drivers justifies foregoing the opportunity to save more unbelted drivers. The reasons for this policy choice are similar to those for depowering passenger air bags.

First, the principle of not affirmatively causing harm when harm would not otherwise occur applies to all vehicle occupants. While it is probably unavoidable that some safety devices may on occasion substitute one type of injury for another type that would occur in their absence, it is not acceptable that safety devices cause a significant number of fatalities in circumstances in which fatal or serious injury would not otherwise occur.

Second, the drivers who might lose benefits as a result of depowering are unbelted drivers. They can protect themselves by taking the simple step of buckling their safety belts as required by the laws of 49 States and the District of Columbia.

Nevertheless, as noted above, due to the possibility of adverse safety tradeoffs, NHTSA is seeking to limit the duration of the tradeoffs by proposing to make only a temporary change in Standard No. 208 to permit or facilitate the depowering of air bags. The agency's planned proposal to require smart air bags would not only enable the agency to make depowering a temporary measure should the adverse tradeoffs actually materialize, but would also ensure that the problem of adverse effects from air bags is fully addressed, and that air bags achieve their full safety potential.

C. Alternative Proposals.

The preceding sections of this notice discuss the benefits of depowering passenger and driver air bags by various amounts, and the net effects on safety. While the agency recognizes that depowering air bags may result in some adverse safety tradeoffs, primarily to unbelted teenage and adult occupants, it believes that depowering represents a desirable temporary means of addressing the problem of fatalities and injuries from air bags.

Having tentatively decided that depowering of air bags is desirable, it is necessary for the agency to determine whether a regulatory change is needed to permit this action and, if so, what the most appropriate change would be.

Manufacturers have asserted that a regulatory change is needed because if air bags were depowered to an appropriate extent, manufacturers would be unable to certify that all of their vehicles comply with Standard No. 208's unbelted test requirements.

As discussed in the PRE, the agency's testing shows that an average 20 to 35 percent depowering of passenger air bags would result in chest g's for some vehicles approaching or slightly exceeding Standard No. 208's 60 g limit for the unbelted test. This indicates that a regulatory change would be needed to permit this level of depowering for these vehicles. The agency's limited data suggest that the standard's other requirements would not preclude this level of depowering, although the 1000 HIC limit would prevent significantly higher levels of depowering.

NHTSA does not have data concerning whether a regulatory change would be needed to permit 20 to 35 percent depowering of driver air bags, but is requesting commenters to provide such data. As discussed in the PRE, when driver air bags depowered to that extent were tested by NHTSA at 30 mph, unbelted chest g's increased from 49to 51. Ford modeling for driver air bags shows similar results, with chest g's rising by only 2 or 3 g's for belted and unbelted drivers. Available NHTSA modeling shows variable results (some chest g's going up and others down), but all were well within the standard at 30 mph. The agency believes that energy absorbing steering columns explain why the driver air bag can be depowered without significantly affecting chest g's. However, the agency conducted only limited testing and did not conduct any angle tests. The agency requests comments, including data, concerning how depowering driver air bags by various percentages would affect the manufacturers' ability to certify compliance with Standard No. 208.

The agency is proposing the adoption of either, or both of two potential changes as alternative temporary amendments to Standard No. 208: either increasing the current chest acceleration limit to 80 g's, or replacing the unbelted crash test requirement with a sled test protocol incorporating a standardized crash pulse. If the agency were to adopt both of these changes, a manufacturer could select either alternative at its option. However, a manufacturer could not mix the two options, i.e., the 80 g chest acceleration limit would not apply in the case of the generic sled test.

A discussion of each of the two alternative approaches being proposed by the agency is presented in the next two sections.

1. Approach I--Temporary Change in Unbelted Chest Acceleration Requirement.

NHTSA believes that the simplest regulatory change would be to amend the requirement which appears to be the factor limiting the vehicle manufacturers' ability to depower current air bags by 20 to 35 percent. This points to reducing the stringency of the unbelted chest acceleration requirement. The agency is proposing to increase the current limit from 60 g's to 80 g's. However, the agency is requesting comments on both higher and lower values, and could select a different value for the final rule.

This alternative has other advantages in addition to its simplicity. Occupant protection would continue to be measured in full-scale vehicle tests, protection in impacts at a range of angles would be ensured, and the other injury criteria would not change. The agency notes that recent biomechanical data generated for NHTSA suggests that, with respect to potential chest injuries, the human tolerance to acceleration is higher for air bags than for belts, because the air bag delivers a more broadly distributed, uniform loading to the chest than does a safety belt. Therefore, an 80 g requirement for occupants protected by air bags appears to be at least as protective as a 60 g requirement for belted occupants.

The agency notes that amending the standard to allow chest accelerations of 80 g's does not mean that chest g measurements in crash tests would necessarily rise to that level. The agency's test data suggest that while a change to 80 g's would be sufficient to permit or facilitate 20 to 35 percent downloading, air bags with progressively higher levels of downloading (beyond 20 to 35 percent) are likely to exceed Standard No. 208's head injury criterion before they exceed the 80 g requirement.

NHTSA also notes that the PRE's estimates of safety impacts for the 80 g alternative do not assume an increase to 80 g's, or to any particular level below 80 g's. The estimates are instead based on the agency's analysis of the effects of depowering air bags by 20 to 35 percent.

The agency's analysis assumes, based on limited vehicle testing,that chest g's would rise by an average of approximately 11 g's for the unbelted 50th percentile male. Since compliance data show that chest g's for this test currently average about 43 g's, the assumed 11 g increase means that the average would increase to about 54 g's for the 50th percentile male dummy.

NHTSA intends for any regulatory change to Standard No. 208 to permit or facilitate quick depowering of air bags. In order to reduce the leadtime for depowered air bags, the agency is proposing, as part of its 80 g proposal, to establish a special two-year enforcement policy for Standard No. 208's unbelted test requirements.

The agency recognizes that, under ordinary circumstances, manufacturers making air bag design changes typically conduct extensive testing to ensure that a vehicle will continue to meet the standard's performance requirements at any particular level. They do so despite the existence of various provisions of Standard No. 208 that provide that "a vehicle shall not be deemed to be in noncompliance with this standard if its manufacturer establishes that it did not have reason to know in the exercise of due care that such vehicle is not in conformity with the requirement of this standard." See, e.g., S4.1.5.3.

While NHTSA generally considers some degree of testing to be necessary to satisfy this "due care" requirement, under the proposed two-year policy, the agency would consider engineering analyses indicating that a vehicle will pass the unbelted test requirements with a depowered air bag as sufficient during that period to establish that the vehicle's manufacturer exercised due care to ensure that the vehicle conforms with the requirement, even in the absence of confirming crash testing. Of course, the agency would retain the right to enforce the requirements of the standard if the noncompliance was due to quality control deficiencies or other manufacturing problems. This policy would be reflected in an appendix to the standard.

2. Approach II--Temporary Replacement of Unbelted Crash Test Requirement with a Sled Test Protocol Incorporating a Standardized Crash Pulse.

In August 1996, AAMA submitted a petition for rulemaking requesting, among other things, an immediate amendment to the requirements for testing the ability of air bags to protect unbelted occupants. The current requirement measures occupant protection in a full scale crash test in which a vehicle, equipped with test dummies at the outside front seating positions, is crashed into a barrier. Specified injury criteria, measured on the test dummies, must be met in barrier crashes at speeds up to 30 mph, and a range of angles up to 30 degrees off-center.

AAMA requested that this crash test requirement be replaced with a sled test protocol. Under that protocol, all of a vehicle, or a portion of the vehicle representing the interior, would be mounted on a sled. The sled would be decelerated from 30 mph according to a standard formula, called a crash pulse. There would not be an angle test, only a direct frontal test. NHTSA notes that sled tests can be used by researchers to simulate what will happen to occupants in real world crashes. The crash pulse for a given sled test is a major determinant of the stringency of the test, and how representative the test is of how a particular vehicle will perform in particular kinds of real world crashes.

To explain further, the term "crash pulse" is defined as the acceleration-time history of the occupant compartment of a vehicle during a crash. This is typically represented in terms of g's of acceleration plotted against time in milliseconds (1/1000 second). Generally speaking, the occupant undergoes greater forces due to secondary collisions with the vehicle interior and restraint systems if the crash pulse g's are higher at the peak, or the duration of the crash pulse is shorter, which would lead to higher overall average g levels.

The crash pulse experienced by a particular vehicle will obviously differ substantially in different types of crashes, e.g., if the vehicle crashes into a rigid stone wall vs. a stack of hay. Similarly, vehicles with different designs typically experience substantially different crash pulses in the same kind of crash, depending on such things as the stiffness of the vehicle structure and amount of crush space. Large cars typically have relatively mild crash pulses, while small cars and utility vehicles typically have more severe crash pulses.

Under AAMA's recommended amendment, the same crash pulse would be used for all vehicles. The petitioner argued that the standard's current test protocol "directly dictates the level of the air bag's inflator power and it is the level of inflator power that unnecessarily increases the risk of injury to vehicle occupants during air bag deployment." AAMA asserted that its recommended test protocol would allow for lower powered inflators to be introduced into the market as quickly as possible while maintaining air bag protection for all occupants.

In its August 1996 petition, AAMA provided the parameters for its recommended pulse along with a suggested mathematical formula, called a sine pulse. The sine pulse suggested by AAMA is described by the mathematical function: A = 15 sin (t/143) Gs.

After examining the sled test protocol initially advocated by AAMA, NHTSA concluded that the standardized sled pulse suggested in the petition is representative of a very soft, or benign crash. Indeed, the agency wondered whether the pulse were so benign that a vehicle could meet the requirements for protecting an unbelted dummy without an air bag.

To answer this question, NHTSA tested a 1993 Taurus according to the sled test protocol recommended by AAMA, i.e., the 143 millisecond (msec) sled pulse (15 g peak). The vehicle did not have a passenger air bag. Although the vehicle had a driver air bag, it was deactivated so that it would not deploy. Although protected by neither safety belts nor air bags, neither of the dummies had responses that exceeded the injury criteria specified in Standard No. 208.

In its November 13, 1996 letter, AAMA suggested that the agency use a more severe crash pulse, 125 msec., which corresponds to 17.1 g. AAMA also argued that the agency should consider injury measurements for the neck in evaluating the crash pulse, rather than focusing solely on whether vehicles without air bags could pass the current Standard No. 208 injury criteria (HIC, chest and femur loads) in a test using the pulse. AAMA indicated that a vehicle could not meet appropriate neck injury assessment reference values (IARV's) in a test using the pulse without an air bag.

NHTSA notes that the revised AAMA recommended crash pulse is similar to that experienced by a large car in a Standard No. 208 test, but milder than that experienced by a typical small car, utility vehicle, or light truck. The PRE provides additional information about crash pulses.

There are potential advantages and disadvantages to the approach of using a standardized crash pulse representative of a large car as a temporary means of addressing air bag fatalities to children. The approach provides maximum flexibility to manufacturers in addressing these fatalities. In its 1984 rulemaking establishing the automatic protection requirements that were in effect until the implementation of ISTEA, NHTSA recognized that technical problems existed in designing air bags that would not pose a danger to unrestrained small children in small cars. Because the crash pulse of small cars is much more severe than that of large cars, more aggressive air bags are needed to meet the standard's injury criteria. The agency stated:

Manufacturers claim that little development work has been done with air bags for small (e.g., subcompact or smaller) cars and that a particular problem in these vehicles is how to protect small children, who are not properly restrained, from the more rapidly deploying air cushion in such vehicles. The Department believes that this problem can be mitigated and that technical solutions are available, as described in the FRIA. However, the lack of experience in this area, as well as the lack of experience for some companies in any form of air bag development, make the Department reluctant to mandate across-the-board air bags. 49 Fed. Reg. 29001, July 17, 1984; See July 11, 1984 FRIA, pp. III-7 to 11.

The AAMA recommended sled test approach would essentially permit the auto manufacturers to use air bags for small cars and other vehicles with severe crash pulses (e.g., utility vehicles and trucks) that are similar to the ones they use for large cars. This would eliminate some of the problems that exist in designing air bags for these vehicles that are not aggressive to children, i.e., the risk of aggressivity would be normalized for all vehicles.

Another advantage of a sled test approach is that it reduces the time and cost of doing certification testing, since sled tests are less destructive of the vehicle. Further, many more sled tests can be conducted in the same time period, since the motor vehicle industry and its suppliers have substantially greater capacity to conduct sled tests than barrier tests.

The primary disadvantage of using a standardized crash pulse representative of a large car is that the test will be less representative of actual performance for small cars and other vehicles with severe crash pulses, i.e., the test measures only air bag performance and not total vehicle performance. The approach also eliminates the effect of angle test requirements, which ensure protection in frontal impacts that occur at a range of angles rather than purely head-on. However, given that recent NHTSA analyses indicate that current fatality reducing benefits of air bags drop off rapidly as crashes diverge from direct "head-on" collisions, deleting the requirement for meeting injury criteria in a 30 degree test might not substantially degrade the "real world" benefits of air bags in such crash configurations. ("Fatality Reduction by Air Bags, Analyses of Accident Data through early 1996," August 1996 NHTSA Technical Report, DOT HS 808 470) NHTSA requests comments on this issue.

As a practical matter, the AAMA recommended sled test approach appears to permit more depowering than the 80 g approach. Under the 80 g approach, Standard No. 208's HIC requirement appears to preclude depowering much beyond the 20 to 35 percent range. The agency does not know how much depowering would be permitted by the AAMA approach, but believes it could be considerably greater than 35 percent, at least for vehicles that currently experience a severe crash pulse in the current Standard No. 208 test. While this maximizes manufacturer flexibility in addressing the fatalities to children, it also raises the possibility of greater adverse safety tradeoffs, especially to unbelted teenage and adult occupants.

In the context of a temporary amendment to Standard No. 208, however, the agency believes it is important to distinguish between what the manufacturers might technically be permitted to do and the actions they would actually take in response to a regulatory change. Because of the substantial differences among current air bags, it is likely that very different levels of depowering are needed for different air bags in order to significantly reduce the risk of child fatalities. For some air bags, 10 percent depowering may be necessary; for others, 60 percent depowering may be necessary.

Because the same standards apply to all vehicles, it is possible that any regulatory change that would permit 60 percent depowering of the most aggressive air bags would permit greater than optimal depowering of other air bags. That does not mean, however, that manufacturers would depower all air bags to the maximum extent permitted by the amendment. Instead, the agency anticipates that the manufacturers would only depower particular air bags to the extent needed to address the child fatality problem, and preserve unbelted occupant protection to the maximum extent possible.

As part of proposing the AAMA recommended sled test approach, the agency is proposing to add neck injury criteria for the 50th percentile male dummy. As indicated above, AAMA argued that the agency should consider injury measurements for the neck in evaluating the crash pulse. The source of the proposed neck criteria is "Anthropomorphic Dummies for Crash and Escape Systems," AGARD Conference Proceedings of NATO, July 1996, AGARD-AR-330. A copy of the relevant pages is being placed in the docket. The agency notes that GM uses the same neck criteria for its IARVs. Data provided by AAMA indicate that, in general, all of these neck criteria could not be met without an air bag.

The proposed neck injury criteria represent peak values for very short duration loading. Much lower loads can be tolerated for longer duration loading. Time dependency criteria may need to be specified. The agency solicits comments on this subject.

The agency is proposing a test procedure similar to that presented in AAMA's petition. NHTSA notes that the proposed procedure specifies that the vehicle, or "a sufficient portion of the vehicle to be representative of the vehicle structu