Michael Paine, Vehicle Design and Research Pty Limited mpaine@tpgi.com.au

Michael Griffiths, Road Safety Solutions michael3@ozemail.com.au

Julie Brown, road safety consultant brownj@planet.net.au
 

Presented at 'Recent developments in child restraint design, installation and regulation'

Monash University Accident Research Centre and RACV, 5 February 2001. Updated 7 Feb 2001

The views expressed are those of the authors and do not represent the views or policy of any organisation.
 

Contents

Abstract

The child restraint designs used in Australia have been shown to provide exceptional protection to child occupants in severe crashes. Cases of serious injury are likely to involve misuse of the child restraint.

Observational surveys of unoccupied child seats indicate that, in NSW, about one quarter of child restraints are incorrectly fitted to vehicle. The most prevalent problem with child seats is incorrect threading of the adult seat belt but in most cases this is a minor problem.

Dynamic tests of child restraints using sleds are conducted for Australian Standard compliance and the Child Restraint Evaluation Program.

Child restraints are now included in new vehicle crash tests conducted under the Australian New Car Assessment Program.

Recommendations are given to enable further improvement of Australian child restraints.

Introduction

This paper discusses methods of assessing the performance of child restraints (CRs) in Australia.

Since the 1970s dynamic testing of child restraints has been required under the Australian Standard for child restraints.

In 1994 New South Wales introduced the Child Restraint Evaluation Program (CREP) to provide a guide to consumers. This program includes sled tests of child restraints.

In 1999 Australian New Car Assessment Program (ANCAP) aligned its testing procedures with those of EuroNCAP. Offset frontal and side impact crash tests of vehicles now include child restraints and dummies in the rear seat.

In addition to dynamic testing the NSW Roads and Traffic Authority (RTA) occasionally conducts surveys of child restraint usage. In 1998 Vehicle Design and Research P/L was contracted by the RTA to conduct a pilot study involving interviews with motorists who had children in the vehicle.

In-depth investigations of real world crashes have been conducted in NSW from time to time. In 1993 a major study of 131 crashes involving 247 children was conducted for the Child Accident Prevention Foundation of Australia (CAPFA - now Kidsafe). In the late 1990s the NSW RTA undertook a "Crashed Vehicles Study" that collected a wide range of data about crash involved vehicles. That study included inspections of 4426 vehicles involved in 2705 crashes. Data from that study are still being analysed but it is expected that it could provide information on several hundred crashes involving child occupants.

New Car Assessment Program

NCAP assesses the crashworthiness of new vehicles and provide a star rating for the protection provided to front seat occupants. Two types of crash test are used in the assessment - a offset frontal crash test and a side impact crash test.

The offset frontal crash is conducted at 64km/h. The vehicle hits a crushable aluminium honeycomb barrier and the crash forces are concentrated on the driver’s half of the vehicle. The side impact involves a moving barrier, fitted with a crushable aluminium front, hitting the driver’s side of the car at 50km/h.

ANCAP crash test procedures are based on those used by the European New Car Assessment Program (EuroNCAP). Under these procedures two child restraints are installed in the rear seat of the vehicle. The child dummies used are TNO P1.5 and P3, simulating 18 month and 3 year old children respectively. In the offset frontal crash the P3 sits behind the driver and the P1.5 sits behind the front passenger. The positions are swapped for the side impact crash test (which does not have a front passenger dummy).

The child dummies are instrumented with head and chest accelerometers. Dummy movement is recorded on high speed film and is analysed to estimate the movement of each dummy.

The child restraint assessment is not included in the vehicle overall star rating but is reported separately by EuroNCAP. Due to the fundamentally different design of child restraints in Australia ANCAP does not currently report the results of child restraint performance (discussed in more detail below).

EuroNCAP child restraint assessment

The EuroNCAP protocol has separate methods for assessing the offset and side impact crash tests.

Offset frontal crash test

Head

To obtain a ‘good’ result the P1.5 dummy head upward acceleration (3ms Z) should not exceed 20g. No limit applies to the P3 dummy.

A ‘poor’ result is obtained if the head excursion (relative to the CR point) exceeds 550mm with either dummy. This is for forward facing CRs. The limit is 600mm for rearward facing CRs. It is a little difficult assessing this head excursion due to the effect of parallax and the difficulty of obtaining good camera angles. However, with top tethers, none of the Australian CRs appear to be allowing anywhere near this large amount of head movement.

Chest

To obtain a ‘good’ result the 3ms resultant chest acceleration should not exceed 41g and the vertical component should not exceed 23g (3ms) . This applies to both dummies.

A ‘poor’ chest result is obtained if 3ms resultant chest acceleration exceeds 55g or the vertical component exceeds 30g (3ms). This applies to both dummies.

Side impact crash test

The protocol notes that "in the absence of satisfactory child dummy and bio-mechanical criteria for side impact, the criteria chosen have been kept simple"!

Head

3ms resultant head acceleration should not exceed 80g.

A somewhat subjective limit is applied to head containment: "No part of the head shall pass outside the forward projected exterior surface of the child restraint". It is difficult to assess this requirement given the awkward video angles that are available, particularly with the EuroNCAP videos. ANCAP uses two onboard cameras that give a much better view of the child dummies than the EuroNCAP videos but analysis is still difficult.

Australian Standards

Child restraints used in Australia must comply with AS1754. The dynamic testing for CRs is set out in AS3629.1. This specifies sled tests for child seats (type B restraints) as follows:

• a frontal impact at about 49km/h with a peak deceleration of 24g and
• a 90 degree side impact test with a peak deceleration of 14g and an impact speed of 32km/h.
• a rear impact test with a peak deceleration of 14g and an impact speed of 32km/h.

For the frontal test AS1754 limits the 3ms resultant head acceleration to 150g. Systems are also assessed for:

• retention of the CR
• retention of the dummy
• separation of load bearing components
• fragmentation of rigid components
• adjuster slip

In addition, an inverted test at 16km/h to simulate a rollover crash is conducted for infant capsules and rearward facing child seats.

Child Restraint Evaluation Program

The Child Restraint Evaluation Program (CREP) is operated by the NSW RTA, NRMA and RACV in association with the Australian Consumers Association. The outcome is a buyers guide to CRs. The assessments are based on the Australian Standard but involve higher crash forces and additional test procedures. In addition to the AS1754 tests described above a frontal test at 56km/h and 34g is conducted.

The side and rear impact tests are conducted at the same speed as the AS but, in the side impact test, a structure that is intended to replicate the interior of a side door is added to the test configuration.

A side impact test is also conducted at an impact angle of 45 degrees.

With child seats a P6 dummy is used for the frontal test and a P3/4 for the other tests.
 
 

Assessment Criteria

The CREP assessment criteria include those covered under AS1754. The following assessments are made with child seats (Type B):

• Head acceleration
• Harness strap forces (frontal test)
• Tether forces, harness forces and seat belt forces (frontal test)
• Head displacement (frontal test) – including rebound – limits apply to upward and rearward excursion (during rebound) but not to forward excursion.
• Head retention (containment) – side impact tests
• Retention of device and dummy
• Adjuster slip
• Buckle release force (frontal tests)

With most criteria there are no limits set for performance – the models of restraint are simply ranked in order of measured values and good performers tend to stand out in these lists. There are specific reasons for excluding child restraints from the ’preferred buy’ list:

• Head excursion outside prescribed limits in frontal test or rear impact test
• Head contact with test rig during side impact test

Comparison of methods

Frontal impact

For child seats (Type B) the Standard and CREP use a P6 dummy in the frontal sled tests. EuroNCAP/ANCAP uses P1.5 and P3 dummies.

The peak sled deceleration and delta Vs are 24g / 49km/h in the Standard and 34g / 56km/h in CREP. Typical vehicle body shell decelerations are around 30g to 40g in the EuroNCAP crash test at 64km/h (delta V about 60km/h).

Analysis of several EuroNCAP tests revealed that, although peak decelerations were similar, dummy injury measurements were substantially less in the NCAP tests than those of the CREP tests. Some reasons for the differences could be

• The larger dummy size used in the sled tests
• The absence of a top tether in the EuroNCAP test. However, early experience with the ANCAP tests involving top tethers suggests that injury measurements are also much lower than those in the sled tests.
• "Stiffer" deceleration pulse with the sled test
• Differences in seat characteristics
• Differences in seat belt geometry.

Side impact

For child seats (Type B) the Standard and CREP use a P3/4 dummy in the side impact tests. EuroNCAP/ANCAP uses P1.5 and P3 dummies.

The peak sled deceleration and delta Vs are 14g / 32km/h in the Standard and CREP. Typical vehicle body shell decelerations are around 20g in the EuroNCAP crash test at 50km/h (delta V about 40km/h).

Again the NCAP injury measurements tend to be substantially less than those of the sled tests.

Real world data in NSW

The CAPFA Study

In 1993 an in-depth study of children involved in car crashes in NSW was initiated by the RTA (the "CAPFA study"). The study was supervised by Dr Michael Henderson. It covered 131 crashes involving 247 children.

There were 38 children in forward-facing child seats. There was one fatality and it involved gross misuse (use of an adult seat instead of the built-in harness). There were minor injuries in 17 cases and five cases where an individual injury was AIS 2 or more. All but one of the serious injuries involved serious misuse of the child seat that included failure to fasten the top tether. Note that "misuse", particularly minor problems, can be difficult to determine with this type of post-crash investigation.

Due to the method of selection and the small sample size, the CAPFA accident cases might not be representative of the general population, but they indicate the importance of correctly fitting and adjusting the restraint.

In a 1996 AAAM paper concerning the CAPFA study Henderson described the severe crashes in which restrained children survived with no serious injuries. The paper concluded: "There are few safety devices that are as effective as child restraints. We found in our study that the only injuries caused by deceleration alone were bruising and abrasion from loads imparted from harness and seat belt webbing. The head remains the most important part of the body to be protected. The principal threat to the restrained child is from invasion of the child's space through impact intrusion, collapsing seat backs, flying glass and loose objects. The child is also at risk if allowed to move out of its space and restraint design should place a high priority on the minimisation of excursion of the upper body in order to prevent head contact."

We recently conducted some further analysis of the CAPFA study data to assess the relevance of the sled and crash tests. This is subject to the caution that the study was biased towards the more serious crashes.

Frontal crashes

There were a total of 18 cases of frontal crashes involving children restrained in child seats. In 16 of these the child was uninjured or only suffered minor injuries, even though in six of these cases the estimated Delta-V exceeded 56km/h.

Side impact crashes

There were a total of 14 cases of side impact with children restrained in child seats. All children were uninjured or only suffered minor injuries. In four cases the estimated Delta-V exceeded 32km/h.

Rear impact crashes

There were a total of 5 cases of rear impact involving children restrained in a child seat. All children were uninjured.

Crashed Vehicles Study

Between May 1995 and June 1998 teams of RTA inspectors conducted inspections of some 4426 vehicles that had been involved in a total of 2705 crashes. Details of the crashes and resulting inspections are still being analysed by the RTA. A total of 2927 cars were inspected. Based on RTA roadside surveys it can be expected that roughly 10% of these would have had a child occupant in a child seat - that is an estimated 270 cases. At this stage the data for these crashes has not been analysed. Furthermore comprehensive injury data, other than that recorded by the Police, has not been collected, although this was intended at the start of the study.

Restraint Fitting Stations

Since 1986 New South Wales has operated a network of authorised Restraint Fitting Stations (RFS). These are small businesses, usually associated with mechanical repair shops, which provide a service fitting child restraint anchorages, installing child restraints in vehicles and advising carers how the child restraints should be used.
 
 

Child Restraint Surveys

Between 1979 and 1989 the RTA conducted surveys of child restraints fitted to cars in shopping centre car parks. The restraints were observed through the windows of parked vehicles to establish whether they had been fitted in accordance with manufacturer's instructions.

Approximately one third of child restraints were found to be incorrectly installed in the 1984 and 1989 surveys. The use of unapproved restraints was checked in the carpark surveys but was not found to be a problem.

Seat belt usage surveys, mostly involving the observation of vehicles waiting at traffic lights, have been regularly conducted by the RTA since 1970. These surveys have included the observation of child occupants and the use of child restraints. The survey techniques do not facilitate the checking of the correct installation of child restraints in the same manner as the car park surveys.

In 1998 the RTA commissioned two types of survey: a repeat of the car parks surveys conducted earlier and a pilot study involving interviews with motorists attending family restaurants and day care centres.

Carpark surveys were carried out at a total 18 locations, including 6 locations outside the Sydney metropolitan area. A total of 1822 child restraints were observed. In a substantial number of cases of child seats and infant capsules the quality of installation could not be determined, usually due to the presence of dark window tinting. Of the 1,177 cases where installation quality could be determined 27% of infant capsules were found to be incorrectly installed. 13% of convertible child seats (which can be used as a rearward facing infant restraints or forward facing child seat) and 27% of child seats were also incorrectly installed. Child harnesses and booster seats had relatively few installation problems, except for 5 cases where a booster seat was used in conjunction with a lap-only seat belt - this is a highly undesirable combination.

Lack of a top tether was a problem in 10% of infant capsule installations and 3% of child seat installations. Note that it is more difficult to establish that a top tether is not being used in the case of infant capsules, compared with child seats, because the forward part of the tether is usually disconnected in order to take the child out of the capsule. However, most capsule designs have a quick release buckle within easy reach and it is most unusual for the carer to unclip the top tether from the anchorage point. Most of the cases recorded top tether problems were where there was no tether attached to the anchorage.

Incorrect threading of the adult seat belt was a problem with 5% of infant capsules and 12% of child seat installations.

The interview/observation survey was a pilot study. A total of 149 interviews were conducted at 12 sites in 10 towns, including 5 towns outside the Sydney metropolitan area. Ten sites were at McDonalds Family Restaurants and two sites were at day care centres. Overall 88% of those approached agreed to both the interview and the in-vehicle inspection of child restraints. Installation quality was determined for a total of 162 child restraints and involved the technician entering the vehicle and making physical measurements of the unoccupied child restraint. Poor installations were found in 27% of the 11 infant capsules, 27% of the 56 convertibles and 30% of the 91 child seats.

Subject to caution about the small sample sizes, installation quality was more likely to be poor if the carer was aged 45 or more (10 out of 13 cases, compared with 23 out of 97 cases for carers aged 25 to 34). This might indicate a problem with grandparents looking after children on weekdays and, possibly, swapping CRs between vehicles.

Restraints claimed to be installed by Restraint Fitting Stations were more likely to be good and had the lowest proportion of poor installations (16% of 40 restraints). However, in all cases of poor installation the CR was fitted by the RFS more than one month previously so we cannot be sure that the RFS installed the CR incorrectly.

Technically the interview surveys were considered to be successful but they were found to be relatively resource intensive. It was concluded that future surveys should endeavour to observe the manner in which the child is placed into the restraint.

The research report on this project has not yet been released by the RTA.
 

Discussion

Top tethers, as used on all Australian child restraints, are extremely effective at limiting forward head excursion - considered to be the most hazardous feature of child kinematics in a frontal crash. Australian crash experience shows that children correctly restrained in child seats can withstand severe crash forces without serious head, neck or chest injuries. Unless there is rearward movement of the front seats, or other intrusion into the child occupant survival space, it is unlikely that a head contact would occur, other than with a restraint component.

This suggests that the EuroNCAP injury criteria for the frontal crash are inappropriate for the assessment of child restraints with top tethers. In particular the limits on head and chest decelerations are considered to be too low and might encourage excessively flexible installations that could result in greater risk of injury from head contacts. For the time being ANCAP is not publishing the results of child dummy injury measurements in cases where a top tether is used.

There are a few cases where near-identical vehicles have been tested by both EuroNCAP and ANCAP (possibly the Daihatsu Sirion, Hyundai Accent and Toyota Camry). Although there were differences for front occupant protection in these cases (mostly airbag differences) it might be useful to compare the performance of the child restraints since the ANCAP tests used child restraints with top tethers and the EuroNCAP cases did not have top tethers.

The safety performance being evaluated by EuroNCAP tends to be associated with the design of the CR rather than the vehicle (the vehicle manufacturer is given the option of nominating a child restraint to be used in the test). There are, however, some vehicle design issues which can influence the protection of child occupants.

Vehicle design issues

Key features of vehicle design that affect the performance of a child restraint are:

Location of anchorages for top tethers. Given the demanding test loads required under the regulations this is mainly an ease of access and adjustment issue. Locations which require the use of a top tether extension strap should be discouraged. Also top tether straps that could be affected by movement of luggage are undesirable.

Retention of the CR by the adult seat belt. This is also an ease of use issue. Lockable retractors for rear seat belts are sometimes used in Europe and the USA to improve the task of installing a child restraint. With top tethers these are not necessary and retractors do not appear to cause problems in Australia.

Rear seat design. A compliant seat back and seat cushion will (in Australia) allow extra movement of the lower part of the child restraint. This is mainly a concern in side impacts where the compliant seat may allow excessive yaw motion of the CR and expose the occupant to direct contact with the (intruding) interior of the vehicle. Compliant seat cushions can also contribute to undesirable rebound motion in frontal crashes. A test methodology for assessing the installation of child restraints was developed for the 1998 RTA survey. In particular, the following should be considered for ANCAP assessments: transverse movement of top of child seat and yaw rotation of child seat (about a vertical axis). Additional items to consider are:

Intrusion of vehicle components into the child's survival space. In particular, front seats which rock excessively backwards are a concern. This may occur during rebound. Another possible source is dynamic buckling of the floor pan under the front seat, causing the seat to rock backwards, even though the occupants are thrown forwards at this time. This was observed in a recent ANCAP crash test.

Provision of placards, stickers and owner's manual instructions that go beyond he minimum required under the ADR 34 for informing vehicle owners about the use of child restraints. For example, the vehicle manufacturer could provide a list of models of child restraint that have been adequately installed in the vehicle. A video showing correct installation procedures would be very helpful.

Conclusions

The child restraint designs used in Australia have been shown to provide exceptional protection to child occupants in severe crashes. Cases of serious injury are likely to involve misuse of the child restraint.

Observational surveys of unoccupied child seats indicate that, in NSW, about one quarter of child restraints are incorrectly fitted to vehicles. The most prevalent problem with child seats is incorrect threading of the adult seat belt but in most cases this is a minor problem. These surveys do not assess the manner in which the child is strapped into the restraint. The CAPFA study found that such cases have a similar proportion of misuse to the proportion of poorly installed restraints and further research is evidently needed into this area of misuse.

Dynamic testing of child restraints using sleds is conducted for Australian Standard compliance and the Child Restraint Evaluation Program.

Child restraints are now included in new vehicle crash tests conducted under the Australian New Car Assessment Program.

Our recommendations for ensuring further improvements to Australian child restraints are:

• Continue with CREP and increase public awareness of the Buyer's Guide so that there is greater pressure on CR manufacturers to perform well.
• Continue to include CRs in ANCAP crash tests. Develop assessment criteria appropriate for Australian CRs. Make data available for research purposes. Compare with EuroNCAP tests of similar vehicles to determine the effects of the top tether in Australia.
• During ANCAP testing, assess the ease of installation of CRs in each vehicle and measure the amount of movement present when CR is correctly installed. Encourage vehicle manufacturers to give attention to CRs in vehicle design.
• During CREP and ANCAP testing, assess the potential installation problems that may occur during installation or adjustment of the CR to suit the dummy.
• Conduct pilot observational surveys and carer interviews that include observing the way in which the child is strapped into the CR.
• Identify design improvements to particular brands of restraint or vehicles that could reduce the chances of misuse and discuss, in confidence, with the relevant manufacturer.
• Identify potential improvements to the Australian Standard and CREP procedures for child restraints.
• Identify potential improvements to Australian Design Rule 34 for child restraint anchorage systems in vehicles. In particular, rear seat design should be given some attention, including shape, softness, seat belt geometry, performance of split seats and resistance to luggage impacts. Rearward collapse of front seats is also an issue with some injuries to rear seat occupants.
• Investigate ways to improve the protection provided to children in booster seats.

Brown J and Kelly P (1998) 'Universal anchorage systems for child restraint devices', Proceedings of the Developments in Safer Motor Vehicles Seminar, Parliament House, Sydney, SAE-Australasia/Staysafe, March 1998.

Eby D and Kostyniuk L (1998) 'Use and misuse of child restraint devices in Michigan', Proceedings of the 16th International Technical Conference on the Enhanced Safety of Vehicles, Windsor.

Eby D, Kostyniuk L and Christoff C (1997) Child restraint device use and misuse in Michigan, University of Michigan Transportation Research Unit, Research Report UMTRI-97-36, September 1997.

Henderson M (1994) Children in car crashes, Child Accident Prevention Foundation of Australia (CAPFA now Kidsafe). Includes a comprehensive bibliography.

Henderson M, Brown J and Griffiths M (1996) 'Adult seat belts: how safe are they for children?', Proceedings of the 15th International Technical Conference on the Enhanced Safety of Vehicles, Melbourne.

Henderson M, Brown J and Paine M (1994) 'Injuries to restrained children' Proceedings of the 38th Annual AAAM.

Henderson M, Brown J and Griffiths M (1997) 'Children in adult seat belts and child harnesses: crash sled comparisons of dummy responses', Proceedings of the 2nd Child Occupant Protection Symposium, SAE P-316.

Hummel Th, Lanwieder K, Finkbeiner F and Hell W (1997) 'Injury risks, misuse rates and the effect of misue depending on kind of child restraint system', Proceedings of the 2nd Child Occupant Protection Symposium, SAE P-316.

Isaksson-Hellman I, Jakobsson L, Gustafsson C and Norin H (1997) 'Trends and effects of child restraint systems based on Volvo's Swedish Accident Database', Proceedings of the 2nd Child Occupant Protection Symposium, SAE P-316.

Johansen P and Young J (1985) Surveys of child restraint fitment in vehicles 1977 to 1985, Roads and Traffic Authority Research Note RN5/85, July 1985.

Kelly P, Griffiths M. Booth M, Lemon J, Crothers N and Franks C (1996) 'Child restraint evaluation program', Proceedings of the 15th International Technical Conference on the Enhanced Safety of Vehicles, Melbourne.

Legault F, Stewart D and Dance M (1998) "Towards improved infant restraint system requirements', Proceedings of the 16th International Technical Conference on the Enhanced Safety of Vehicles, Windsor.

Lowne R, Roy P and Paton I (1997) 'A comparison of the performance of dedicated child restraint attachment systems (ISOFIX)', Proceedings of the 2nd Child Occupant Protection Symposium, SAE P-316.

Lumley M (1997) 'Child restraint tether straps - a simple method for increasing safety for children', Proceedings of the 2nd Child Occupant Protection Symposium, SAE P-316.

Lumley M (1998) 'Australian child restraints lead the world', Proceedings of the Developments in Safer Motor Vehicles Seminar, Parliament House, Sydney, March 1998.

Paine M. (1998) Child restraint surveys in New South Wales: 1998, NSW Roads and Traffic Authority Research Report 98/3, November 1998, Road Safety Branch, Sydney.

Paton I, Roy P and Roberts A (1996) 'The frontal impact performance of child restraint systems conforming to the ISOFIX concept', Proceedings of the 15th International Technical Conference on the Enhanced Safety of Vehicles, Melbourne.

Paton I and Roy P (1998) 'Development of a sled side impact test for child restraint systems', Proceedings of the 16th International Technical Conference on the Enhanced Safety of Vehicles, Windsor.

Pedder J, Gane J, Pasco D, Deibert M and Lumley M (1997) 'Usability trials of alternative child restraint attachment systems', Proceedings of the 2nd Child Occupant Protection Symposium, SAE P-316.

Sampson D, Lozzi A, Kelly P and Brown J (1996) 'Effect of harness mounting location on child restraint performance', Proceedings of the 15th International Technical Conference on the Enhanced Safety of Vehicles, Melbourne.

Yann Campbell Hoare Wheeler (1994) Restraint use attitudes and knowledge prior to and following the 1993 Rear Seat Child Restraint Campaign amongst three non-English pseaking background communities, Roads and Traffic Authority Research Note RN 15/94.

Webber K. (2000) 'Crash Protection for Child Passengers: A Review of Best Practice', UMTRI Research Review July-September 2000, Vol.31,No.3

Winston F, Durbin D, Arbogast K, Meaney D, Bents F and Werner J (1998) 'A comprehensive surveillance system to investigate targeted issues in child occupant protection', Proceedings of the 16th International Technical Conference on the Enhanced Safety of Vehicles, Windsor.

Acknowledgements

The paper is derived from a research report prepared for Australian NCAP and the assistance of the ANCAP Technical Committee is gratefully acknowledged.

Harry Vertsonis and David Harkness from NSW RTA assisted with information about the 1998 Child Restraint Survey and Crashed Vehicles Study.

The high quality and proven effectiveness of child restraints in Australia can be attributed, in large part, to decades of work by Paul Kelly who recently retired from the NSW RTA.

Crashlab conducted the sled and crash tests.

Links.