Writing in Melbourne's The Age of May 19, 2000, Jordan Baker quoted figures from a recent Bureau of Transport Economics report showing that the annual bill for road accidents in Australia was running at an estimated $15 billion, a substantial component of this figure being attributed to fatal accidents. In 1999, 1763 people had died in 1547 fatal road accidents, and with the cost to the community of each fatal road accident assessed at approximately $1.7 million, this put the total bill for this class of accident beyond $2 billion.

In 1997, my wife and I - and the driver of another vehicle - came close to being part of that year's road accident statistics. As it happened, the only injuries suffered were seat-belt related, but this was more a matter of good fortune than anything else.

To be personally involved in a road accident is a terrifying and traumatic experience; fortunately for us, the short-term scars soon healed. However, there were consequences of this accident I have still to come to terms with. These concerned the protracted communications and correspondence with various road safety authorities and allied organisations that had followed the accident. The issue? Whether the design of a particular type of traffic light system could be legitimately held responsible for causing the accident.

The drama began when my wife, who was driving at the time, wished to turn right at a signalised T-junction situated on the Burwood Highway dual-carriageway between Upwey and Belgrave in the Lower Dandenongs, east of Melbourne (see schematic). When making a right turn at this intersection one does so under the control of two sets of lights.

The first set (designated system 1), is located on the pedestrian island (PI), the second on the far right hand side (RHS) of the road ie in the direction of turn. As we approached the T-junction, we slotted-in behind a small van, also about to make the turn. It was dark at the time with both sets of lights showing full green discs. As the van commenced its turn ahead of us, so my wife - under the mistaken impression it was safe to do so - followed suit.

We can both recall my remarking, "I hope you know what you're doing", so, as the van's turn opened up our visibility ahead, I must have been aware of the headlights of a car bearing down on us from the opposite direction. The next thing I recall was shouting "No! No! No!" as I saw a collision was imminent.

Our car was hit on the front left side and was spun violently around before coming to a sudden stop. As my seat belt reacted, I was conscious of being shaken around in what seemed my own private cocoon. Since there was scarcely any time for the driver of the other vehicle to reduce speed, the three of us were extremely fortunate to have escaped serious injury. It could certainly have been far, far worse had we been hit amidships. As it was, our car was a 'write-off'.

The police, on arrival, issued an 'on the spot fine' for having failed 'to give way at intersection'. A few hours later, however, I had already begun to have serious doubts about the cause of this 'right through' crash (as crashes of this type are termed).

As both participant and eyewitness to these events, I instinctively felt there was a potentially fatal flaw in the design and operation of the lights. Here we had a situation where two vehicles, entering a junction from different directions, were both under the 'control' of a signalised system showing two full green discs to each vehicle - surely an unsound piece of road traffic engineering, inherently unsafe and a totally unnecessary means of controlling traffic through such intersections. For the unwary - and as we had found out to our cost - it seemed a classic instance of 'an accident waiting to happen'.

On filling out the insurance company's 'motor vehicle claim' form the next morning, in the section marked 'Who was at fault and give reasons', my first reaction had been to write: 'Our fault, but filtering of lights leaves great deal to be desired'.

On returning to Adelaide I felt concerned and motivated enough to conduct a survey of traffic light systems at dual-carriageway/side road intersections in both the metropolitan and central business district (CBD) of Adelaide. Particular attention was to be given to the visual and mode of operation of these lights.

Before discussing the results of this survey, it's as well to go through some of the basic definitions and conventions associated with signalised turns.

Right-turning traffic may be controlled in two ways:

    1 When a full green disc is displayed it signifies to the driver that where there are sufficient gaps in oncoming traffic then a 'filter' turn is permitted. Filter turns result in shorter delays and allows greater flexibility (code for 'judgment') on the part of the driver. Against this, the driver may be exposed to greater risk than when making....

    2 A 'controlled' turn, which is achieved by the use of arrows. A green arrow signifies to the driver that while making the turn, opposing traffic is facing a red light. Conversely, a red arrow tells the driver that no right turn can be made - even in the absence of oncoming traffic . Controlled turns are particularly appropriate for high density traffic situations where there is little opportunity otherwise for drivers to make right turns safely.

As was explained to me, the essence of good traffic-light design and operation is to maximise the volume of traffic through an intersection, at the same time ensuring that safety requirements are not compromised. Inevitably in the real world, there will always be those situations where drivers find themselves frustrated by facing a red arrow - and no opposing traffic. Conversely, there will be those situations where a greater prevalence of controlled turns would be appreciated.

Before I conducted the survey, I had thought there might be T-junction traffic light systems similar to the one we had fallen victim to in Upwey. But of five configurations of signalised T-junctions eventually located, only one, in the CBD, functioned similarly to the system encountered in Upwey. The survey was not seen as exhaustive - for example, unlike some cross-road traffic lights sighted in Adelaide, no signalised T-junction system was found fitted with plain-English, flashing 'turn with care' signs.

A meeting with a Transport-SA road traffic engineer later confirmed that, in the Adelaide metropolitan area, where pedestrian-island lights were fitted, there were NO full green discs fitted on RHS traffic lights of T-junction systems. (For some reason Adelaide CBD lights are not under the control of Transport-SA.) Nor for that matter have any such systems since been sighted when driving in or visiting Sydney, or cities in UK and New Zealand.

In my survey the following T-junction traffic light systems were sighted for traffic turning right from a dual-carriageway:

    System 2 - the most basic - is a standard traffic light system of three full discs positioned on the pedestrian island. In this particular configuration, vehicles wishing to turn right can only do so 'with care' ie other vehicles travelling in the opposite direction through the junction always have priority.

    For system 3, situated in the CBD, right-turning traffic may do so, first, by a full green disc on the RHS light (filtered turn), second, by a green disc/arrow combination (controlled turn). The filter turn signal might not be that hazardous during the day, the speed and volume of oncoming traffic making it unlikely for motorists to misinterpret the lights. However, in quieter periods, eg the early hours of the morning, it could be a different story.

    For system 4 the right-turn is under the control of two 'in-tandem' pedestrian island lights. A vehicle entering the turn-right lane has the capacity to override the duration of the filtered component of this sequence. Visually, system 4 regresses to system 2 during the turn right 'with care' phase.

    System 5 is located along a dual-carriageway where there is an 80km/hour speed limit. Safety considerations being paramount, traffic turns right in 'controlled' mode only.

    Though System 6 is located on a dual carriageway, there is no pedestrian-island traffic light. The RHS shows a full green disc during a filtered turn. Thus there may be some increased potential for the incidence of 'right through' accidents at this particular intersection.

    System 6 is similar to signalised T-junctions located on main roads considered not wide enough to be classed as dual-carriageway.

Where pedestrian island traffic lights are present in metropolitan Adelaide signalised T-junction systems, the convention of not having a full green disc on RHS poles appears to have been strictly followed.

Having conducted the survey, I contacted various Victorian and South Australian transport/road authorities, State politicians and my insurance company with a view to providing feedback on an accident I had had some personal involvement in. I had thought such an approach would be welcome, particularly if it transpired that feedback was a comparatively rare commodity in road accident research. After all, I reasoned, had we been killed, no one would have been any the wiser on why, or how, this accident had occurred.

These discussions and/or correspondence drew a varied response:

    VicRoads intimated that changes might be made to the lights, but my understanding, years later, was that no such changes had been undertaken.

    One engineer sought to deflect any criticism of the system's design by arguing that in addition to our respective ages, our reflexes had been dulled by the 800km drive from Adelaide. and thus driver fatigue was likely to have contributed to the accident. I gained the strong impression he didn't take too kindly to an outsider coming in and criticising traffic light design.

    Likewise, our insurance company wrote that while 'they could understand our concerns at the way we felt the sequence of lights had contributed to the accident, this in no way could be seen as changing the facts - that my wife had failed to give way to oncoming traffic'. Nor were they interested in their solicitors reviewing the issue retrospectively. Having displayed such a blinkered approach to the incident's aftermath, I saw little point in renewing my insurance policy with the company.

Two years after the accident, I sought to expand and formalise the research I'd conducted through the (then) Federal Office of Road Safety (FORS), now the Australian Transport Safety Bureau (ATSB). In 1989 the FORS made grants available for small-scale, road-safety research projects, the maximum single award being set at $20,000. As stated by FORS, the basic objectives of their awards scheme were twofold:

    1 To promote innovative, worthwhile research into road safety and
    2 While applications from established researchers in the field of road safety were welcome, it was hoped to broaden the base of effective road safety researchers by encouraging established scientists, recent post graduates and commercial and community-based organisations to undertake research in the road safety field.

Given the scheme's financial constraints - in 1999 the total grant was $60000 - applicants were informed that 'only 3 or 4 'very competitive' awards would be granted'. Projects were to be assessed against the following criteria:

  •  Scientific merit,
  •  Potential benefit to road safety in Australia,
  •  Cost of the project,
  •  The extent to which the problem stood alone, and
  •  The capabilities and professional standing of the applicant and chief investigator.

Coming from a defence science studies background, I felt comfortable in applying for such a grant, even so I quickly realised the likely cost/benefits of this scheme were likely to be severely compromised by the emphasis placed on 'competitiveness' rather than 'need'.

It seemed to me that throwing a piffling $60,000 at a problem carrying a multi-billion dollar cost tag was a pointless exercise on FORS's part. Would scientists outside the road safety field really feel it worth their while applying, if the great majority were destined to receive rejection slips? Such shortsightedness must surely prove self-defeating to the scheme's objectives and my worst fears were confirmed when of 38 applicants, only four received financial support. I was not amongst them.

Thus, such issues as:

    How many signalised systems are there in Australia of the type we encountered, and where are they located?
    What is the incidence - and cost - of 'right through' crashes at such signalised intersections?
    Is there evidence to suggest that where such crashes occur, they are more likely to involve visitors/tourists to a region rather than local residents?
    Does the incidence of crashes at such signalised T-junctions differ significantly from those at other T-junctions?
    Would a standardised design of traffic light system be likely to result in improved safety at signalised T-junctions?

were destined to remain, as far as I was concerned, 'on hold'.

Long after all the dust had settled I contacted a road safety expert in UK and asked him what he thought of the layout and operation of System 1. To paraphrase his response, he thought that System 1 had little going for it, but don't expect anyone to ever admit that a mistake had been made in its design. Well, he was certainly on the money there!

With no access to precise data on this problem, one can only theorise on the likely accident rate at T-junctions. For example, if the number of vehicles making filtered right turns through that intersection were 500 per day and the chance of a 'right through' crash was found to be 1 for every million turns, then the likelihood of there being one or more crashes over a one year period would only be the order of 17%. If this chance of a crash increases to 1 in 250,000 turns, then it's almost an even money bet there will be one or more accidents per year. Increase the chance of a collision to 1 in 100,000 turns, and there is now an 84% chance of one or more crashes occurring over a one year period.

While the Monash University Accident Research Unit has done much worthwhile work on 'right through' crashes at signalised intersections, I still do not know whether their studies have considered whether the actual design of signalised T-junction systems can be a factor in 'right through' crashes. That I have been unable to find other systems of this type elsewhere in the world over the last seven years - apart from Melbourne and one in the Adelaide CBD - should provide sufficient justification for investigating whether their safety record justifies their removal or not.


Having failed to make any impression with various road safety organisations re- the possibility of there being a problem in this area, I retired to my tent for some five years. What finally tilted the balance in favour of resurrecting the painful memories of some seven years ago was the recent discovery of an apparent anomaly in a metropolitan Adelaide signalised T-junction system.

Over the last few months I have frequently made right turns from the stem of a T-junction located in an outer suburb of Adelaide, close to Flinders University, System A. When turning right, the traffic light in the direction of the turn (given its location, this is the only one the motorist turning right need be concerned with), consists of a single green arrow. signifying that all traffic travelling in other directions must stop while this arrow signal is operating.

Another signalised T-junction system in the Adelaide metropolitan area, System B, shows a full green disc, in addition to the green arrow. Again, this indicates that all other traffic at the intersection must stop while these lights are operating.

However, a third T-junction traffic light, System C, shows just a full green disc. While this light operates, no other traffic may pass through the junction (except there is a written sign stating that the vehicle turning right from the stem must give way to any pedestrian who is crossing the road forming the top of the T-junction). Another system similar to System C - plus its attendant sign - has been sighted elsewhere in the metropolitan area, suggesting that the selection of a full green light, for what is undeniably a controlled turn, is quite deliberate. But to confuse matters further, at the System 3, CBD T- junction referred to previously, traffic making a controlled turn right into North Terrace from Pulteney St does so with a full green light - but in this instance there is no written sign present to indicate that pedestrians have right of way.

I sense there is something of an anomaly in the way Systems 1 and C function. For

System 1, the full green disc in the direction of turn on the right hand side indicates a filtered turn, insofar the driver is only able to turn right 'with care' into the stem. Yet the doubling up of the full green disc on the pedestrian island and the full green disc on the right hand side, could make it all too easy for strangers to fall into the trap of confusing this filter signal as a controlled signal. Such confusion need only be momentary for this to have dire consequences for road safety at such intersections.

Yet, in spite of the need for clear, unambiguous signals, there are still situations such as

System C, where a full green disc in the direction of turn is being used to govern what is obviously intended as a controlled right turn from the stem.

While the set-up for System C does not of itself constitute a road safety hazard, if Australian Road Safety authorities are genuinely concerned to have a standardised, unambiguous set of rules for traffic-light-controlled right turns, would it not be far more sensible to replace System C with either System A or B? In other words, remove the current situation where a full green light can signify a controlled turn in one direction, yet not in another?

A few months ago a fatal 'right through' accident occurred at a signalised, cross-road intersection not that far from where I live. A middle-aged woman had been killed when a bus had struck her car amidships while she was endeavouring to make a right turn from a dual-carriageway. At this particular intersection, traffic turning right is controlled by a traffic light located on the pedestrian island, and chances are the accident occurred during the 'turn right with care' phase ie while the full green disc was showing. Later I contacted the police road-accident branch to determine whether there was any evidence to indicate she had been racing through the turn in a vain attempt to beat the bus. Alternatively had she driven in such a manner to suggest she had been completely oblivious of the approaching bus having the right of way? The police representative was not giving anything away. The accident file had simply been marked 'driver error' and that was that.

But could the accident have been attributable, not so much to a 'misjudgment' on her part, but more a 'mis-interpretion' of what she thought the traffic lights were telling her? In this particular instance, it has to be said, no blame could be attached to the set-up or mode of operation of these lights, nonetheless something had gone horribly wrong, and one might have hoped that crash investigators would have been awake to the possibility of her having mis-interpreted the traffic lights. And, allowing for this possibility, might there not be some way of reducing the chances of such an event recurring at this, and similar interesections? For example, if the installation of flashing, plain-English signs were regarded as prohibitively expensive, would a flashing orange arrow next to the full green light (say) improve the safety of filtered right turns? (I guess some expert will come along and tell me where the flaw is in this design. but for mine such a visual presentation would have got our attention that much faster on the night in question.) Of course in the final analysis, it's a cost-benefit issue: does the implementation of such a change appreciably reduce what is a remote but, nonetheless, finite probability of this particular class of road traffic accident occurring?

But all this aside, my wife and I remain extremely thankful that we survived this traumatic experience and that, courtesy of this medium, it has at least been possible to have had the opportunity of voicing these concerns.

This webpage last updated April 27, 2005

Back to Top of Page

Table of Contents

Back to Main Page

Email me: