Home page

Main topics
News Headlines

Frequently asked Questions
For Policy Makers

Research evidence
Misleading claims
Helmet laws

Search Engine

New Zealand
Other countries

Full index

Policy statement

Oblique impact testing of cycle helmets

Mills NJ, Gilchrist A
Int Journal of Impact Engineering, 2008;35(9):1075-1086

Related paper:
Finite-element analysis of bicycle helmet oblique impacts
Mills NJ, Gilchrist A
Int Journal of Impact Engineering, 2008;35(9):1087-1101
(Mills and Gilchrist, 2008b)

Original paper


Summary of paper (based on authors' abstract)

The performance of bicycle helmets was investigated in oblique impacts with a simulated road surface. The linear and rotational accelerations of a headform, fitted with a compliant scalp and a wig, were measured. The peak rotational accelerations, the order of 5 krad/s² when the tangential velocity component was 4 m/s, were only slightly greater than in comparable direct impact tests. Oblique impact tests were possible on the front lower edge of the helmet, a site commonly struck in crashes, without the headform striking the ‘road’. Data characterizing the frictional response at the road/shell and helmet/head interfaces, were generated for interpretation via FEA modelling.

BHRF Commentary

In the introduction to their study, Mills and Gilchrist cite some previous research which found that most impacts to bicycle helmets in accidents occur at the front and sides of the head and oblique to a surface. High frictional forces can occur if helmets have no shell, and if the surface is rough concrete they can fracture into several pieces. Consequently, it is stated that current helmets all have an external shell. Performance of helmets in oblique impact tests is described. A parallel paper uses some results to validate finite element analysis (FEA) of the tests (Mills and Gilchrist, 2008b).

This study has three important limitations.

The first is that it evades the critical issue of the effect that the addition of a helmet has on rotational acceleration of the head. It merely says that it is difficult to prove that helmets attenuate rotational head accelerations without carrying out comparable tests without a helmet. Not only does it not attempt to do such tests, it makes no reference to other research which did. Corner, Whitney, O'Rourke and Morgan, 1987 found that in impacts to the jaws of cadavers the added weight of a helmet could increase rotational acceleration; its potentially lethal effect was noted. More recently, King et al, 2003 and St Clair and Chinn, 2007 made similar findings. Wearing a helmet is therefore likely to increase the risk of serious injury to the brain in some circumstances. Unless these findings can be shown to be wrong (which Mills and Gilchrist do not), it is inappropriate for governments to compel or even to encourage helmet wearing.

Second, the study is limited to mechanics of non-living systems including simulated heads, helmets and road surfaces. Its only indication that this deals with just a part of the problem of protecting a living brain from injury is to note the criticism of  Henderson, 1995 that solid headforms do not mimic the deformable characteristics of the human head. Ironically, to meet this criticism, tests of helmets containing instrumented cadaver heads are suggested without any reference to Corner, Whitney, O'Rourke and Morgan, 1987. Further, the study appears to lack a sufficient understanding of scientific knowledge of injury to the brain, not even mentioning its main types, focal and diffuse. Henderson also commented that his criticism of solid headforms is especially relevant to children and to pass standard tests padding has to be firmer than might be desirable on theoretical grounds. Corner argued that to meet the needs of children, the density of padding in helmets should be reduced from 50 to 30 kg/m³, but Mills and Gilchrist show no concern about the increase in helmet liner densities over the last 20 years, to compensate for large ventilation slots. They cite densities of 70-100 kg/m³.

Third, the tests simulate conditions of slow speed bicycle-only crashes. Table 4 of the paper specifies a tangential velocity of 3.6m/s and normal velocity of 4.5m/s. This is equivalent to a cyclist who was travelling at 13 km/h falling from a height of one metre – less than the 1.5 metres specified in the EN1078 standard. Because higher speeds at impact are excluded, such as occur in downhill cycling and collisions with a moving motor vehicle, the tests are not representative of most accidents where severe injury to the brain is of concern. As the headform rotational acceleration was rarely greater than 5 krad/s², the study concludes that it was unlikely that any diffuse brain injury would occur if the criteria of rotational acceleration greater than10 krad/s² and rotational velocities greater than100 rad/s are valid. By contrast, Corner found rotational acceleration averaging 58 krad/s² at 45 km/h. The study’s conclusion is therefore of little value and its consequent claim that the criticisms of Curnow and Franklin are invalid is a non-sequitur.

Deficiencies in methods throw other doubts on the study’s conclusions. For example, that the large holes in modern helmets seemed not to affect friction upon impact. However, the test surfaces were essentially flat. In real life, surfaces are more usually uneven. In an examination of 2,272 bicycle accidents in South Australia where the victim reported a head injury, 456 were assessed as having struck a sharp-profile object, such as a kerb or protrusion on a motor vehicle (Somers, 1999) It could reasonably be expected that holes in a helmet encountering such an object, or road metal protruding from tarmac, would cause it suddenly to grip and to increase angular acceleration, despite the study’s suggestion that tarmac is not as rough as concrete. The study also states that the local indentation of the exterior of a helmet with a micro-shell (0.35- 0.6 mm thick) over a high-density EPS liner is relatively small. While this might be so for the conditions of its tests, it might not be for higher speeds and impacts upon uneven surfaces.

The study’s final aim was “to investigate assertions made by anti-helmet campaigners” citing Curnow. As well as it being inappropriate so to describe contributions to the debate in the form of articles in refereed journals, the description is erroneous.

The first error by Mills and Gilchrist is to attribute to Curnow, 2003 the argument that bicycle helmet design reflects a discredited theory that brain injuries are caused by peak linear acceleration. However, Curnow does not use the word ‘peak’. Diffuse axonal injury (DAI), the commonest cause of disability after head injury, including the vegetative state, is produced by angular acceleration at a lower rate and longer duration than that which injures blood vessels. Curnow cites research which suggests that the use of padding in cars and motorcycle helmets might increase the risk of DAI. Insertion of the word ‘peak’ by Mills and Gilchrist would appear to reflect an insufficient understanding of the causes of brain injury.

A second error is to assert that a premise of Curnow's paper is that the majority of bicyclists’ head injuries are due to rotational acceleration and that this can be refuted by statistics of the type of head injuries suffered. In fact, the focus of Curnow is injury to the brain and its mechanisms, rotational acceleration being one, and he cites research on its effects thus:

“Graham et al. (1995) noted that DAI is the commonest cause of disability after head injury, including the vegetative state, and that it occurs mainly in road traffic accidents. In Glasgow, 45 out of 177 patients with fatal non-missile head injury were found to have DAI, judged to be identical to that produced in the subhuman primate by angular acceleration (Adams et al., 1982). In Australia, 29 out of 62 patients fatally injured in traffic accidents had DAI of similar character (Blumbergs et al., 1989) and the brain of a child pedestrian who died after being struck by a car showed injuries associated with angular acceleration (McCaul et al., 1988).”


Corner, Whitney, O'Rourke and Morgan, 1987

Corner JP, Whitney CW, O'Rourke N, Morgan De, 1987. Motorcycle and bicycle protective helmets: requirements resulting from a post crash study and experimental research. Federal Office of Road Safety Report CR55.

Curnow, 2003

Curnow WJ, 2003. The efficacy of bicycle helmets against brain injury. Accident Analysis and Prevention 2003,35:287-292.

Henderson, 1995

Henderson M, 1995. The effectiveness of bicycle helmets: a review. Motor Accidents Authority of NSW .

King et al, 2003

King AI, King H, Yang LZ, Hardy W, Viano DC, 2003. Is head injury caused by linear or angular acceleration?. IRCOBI Conference, Lisbon 2003 .

Mills and Gilchrist, 2008b

Mills NJ, Gilchrist A, 2008. Finite-element analysis of bicycle helmet oblique impacts. Int Journal of Impact Engineering 2008;35(9):1087-1101.

Somers, 1999

Somers RL, . Letter to Federal Treasury 12.10.1999. South Australian Department of Human Services (obtained under FOI request to the Treasury).

St Clair and Chinn, 2007

StClair VJM, Chinn BP, 2007. Assessment of current bicycle helmets for the potential to cause rotational injury. Transport Research Laboratory PPR213.