A bit of a follow-up to last week’s post about the Perne Road roundabout, looking at the potential issues, and what could have been done instead.
This roundabout has now hit the headlines because a child has been injured while cycling on the roundabout, on Wednesday evening. I don’t think it’s massively helpful to leap to conclusions on the basis of one incident, but it’s certainly worth looking at the general design flaws with this roundabout, and the alternative ways in which it could have been designed.
For me, the central problem is that cycling has not being designed for explicitly. Instead, it has been bodged into pedestrian-specific design, and into motor vehicle-specific design, simultaneously. Almost all the potential issues flow from this failure. The roundabout design expects people on bikes to behave like pedestrians, or like cars; something genuine Dutch design would never do.
For a start, the ‘shared use’ paths around the edge are quite obviously footways, on which it is permissible to cycle. They are not cycle tracks, with clearly defined routes. The result is cycling in a pedestrian-specific environment, and this, coupled with a lack of clarity, presents a number of problems.
With ‘shared footways’, drivers will have less certainty over where a cyclist might be heading. Take the scenario below, with the path of a cyclist represented by the blue arrow.
Where is that cyclist going? Across the crossing? Or away from the crossing, along the road?
The driver doesn’t know if the cyclist is, or isn’t, going to use the crossing. The cyclist is travelling across an expanse of tarmac, and their intentions aren’t clear. The driver may assume wrongly.
Contrast this with a Dutch roundabout (in Assen) –
It’s much more obvious to drivers, at an earlier stage, where cyclists are heading, and they can respond accordingly. (Note that on this roundabout, cyclists don’t have priority.)
And the same is true from the perspective of people cycling. They have more time to assess which direction a driver is taking – staying on the roundabout, or leaving it – and therefore will have more opportunities to cross, more safely. Again, this is without cycle priority –
The Cambridge roundabout does not have this cycle-friendly feature. Because the crossing points are not set back any distance from the roundabout, there’s little time in which to assess which way drivers might be heading. In many instances, it may be too ambiguous to take a chance.
Placing the (pedestrian) crossings at these locations close to the roundabout also means they are blocked by drivers queueing to enter the roundabout, rather than left clear, as on a Dutch roundabout, by setting the crossing points back from the perimeter.
Funnily enough, although I’ve criticised the Poynton scheme, this ‘setting back’ of the crossings has been done correctly there, approximately one car length back from the ’roundabouts’.
This means people can cross behind stationary vehicles, rather than trying to cross in front of a vehicle that might be about to jump into the roundabout.
This ‘set back’ design approach also allows drivers to deal with crossing cyclists/pedestrians, and entering/exiting the roundabout, in two separate stages.
To compound these issues of uncertainty about where people are going, drivers have to contend with people cycling on the road, and on the footway, simultaneously, as they enter and exit the roundabout, rather than dealing with cyclists at one clear crossing point, on defined paths. This is a point John Stevenson makes here –
Drivers don’t know where cyclists are going to be. Because cyclists can either use the main carriageway or the shared-use, off-carriageway paths, drivers are expected to look for cyclists in a number of places at each arm of the roundabout, instead of just one.
Unnecessary complication has been added by putting people cycling on two different forms of route across the roundabout.
Another issue John identifies – having visited the site – is that a shared-use footway, by definition, involves mixing up pedestrians and cyclists together, rather than separating them, and that can be an uncomfortable experience for pedestrians, particularly in areas with high levels of footway cycling. Again, this problem is not one that should have been created.
What effect might the narrowed carriageway have on people who continue to cycle on it? John thinks it might make collisions more likely, as people cycling will be closer to motor vehicles (and there also might be a temptation to squeeze through). That said, the geometry has been tightened, which should lead to lower vehicle speeds – so the collisions would probably on balance be less serious. Swings and roundabouts, although it is obviously far too early to make definitive judgements. In any case, a roundabout with this volume of motor traffic shouldn’t – in principle – be designed with the expectation people will be cycling on the carriageway.
Finally, there has been an awful lot of discussion about whether or not a genuine Dutch-inspired roundabout design would offer cyclists priority over motor traffic, or not. To me, that’s not a particularly pressing issue, compared to the overall design problems set out here. A Dutch roundabout with priority would look very similar to a Dutch roundabout without priority. Cyclists would have clear routes, separated from pedestrians – routes which would make it obvious to drivers what they are doing. Likewise the paths that drivers are taking would be clear, and the roundabout would be designed to maximise crossing opportunity. This roundabout achieves none of those outcomes.
My personal inclination – and I’ve been persuaded on this point – is not to offer cyclists priority, for the main reason that it is safer (remember, this is an entirely new kind of treatment for British drivers), and also because the loss of convenience is marginal, if the roundabout is designed properly. We should remember that no Dutch roundabout offered Dutch cyclists priority, at all, until the 1990s, by law. It was only for reasons of convenience – not safety – that his law was changed, and priority was switched in urban areas.
Priorities can be changed easily – bad design can’t.
UPDATE
Just a quick rough-and-ready size comparison between the Perne Road roundabout, and the Dutch example illustrating this post (which is here, by the way). Not an exact science, but there appears to be more space available at Perne Road.
I think one of the main issues is that the size of the roundabout island itself has been maintained at Perne Road – the island in the Dutch example is considerably smaller, about 15m in diameter, compared to ~23m at Perne Road.
I understand that it would have been expensive to ‘shrink’ the roundabout, due to a culvert, but I think maintaining its size might have led to some of the problems and issues detailed here.
As Easy As Riding A Bike 
You may well have a point about the design of the crossing points at this junction – I’ll suspend judgement until I’ve been there. However, to say that crossings should or should not be given priority on a blanket basis is far too simplistic.
Roundabout arms vary hugely in the volume of traffic they carry, and requiring cyclists to give way as a matter of course will mean that they could suffer long and potentially unlimited delay at the busiest points on the network. Hardly the hierarchy of users we would wish to see. You seem to believe that separating cyclists from pedestrians on the paths will make everyone want to use them. They won’t if it takes 3 minutes to cross the junction.
I think you’re overlooking the potential offered by the new Zoucan crossing (there, I’ve named it) which I believe is very strong and will be respected by drivers. Obviously it’s new though, so we will have to see.
Well, I’m not saying that priority should, or shouldn’t be given on a blanket basis!
I’m just saying that the issue of priority isn’t the central one here. A good roundabout design for cycling would look the same, whether or not priority is given on all or any of the arms. If there’s a strategic case for providing priority on some of the arms, then it should be given, even if that means some lowering of safety.
That priority can be easily adjusted, just by changing the markings. Indeed, this is what the Dutch did, on a large scale!
You talk about the “expectation” of cycling behaviour. For me I think the priority would have a large bearing where you might expect to find me. I don’t wear lycra or ride a road bike but I am often running late and lacking patience, and my assumption would be that a roundabout where traffic flows are high and I lose priority on every arm would be less than optimal for directness.
Of course my decision would depend on how busy it was – time of day – and who was with me. But I’m sure it would be politically unwelcome to try and ban cyclists from the carriageway, especially if that means losing priority.
Just a note: there already is light-controlled (pedestrian) crossings on one of the busier arms, which were retained through the roundabout re-design. However, it is substantially off the desire line, being about 6 car lengths in. I assume this is to stop vehicles backing up onto the roundabout.
We will potentially have a cycle zebra in Cambridge on Huntingdon Road. It was part of the Cycle City Ambition project, though it was an optional part of the plans depending on whether DfT permission is granted.
Sorry to hog the comments but thinking about this some more, isn’t this “let cyclists give way, it’s safer” meme a failure to really tackle the problem; which is driver behaviour and compliance?
When a pedestrian is knocked down at a zebra crossing, which is the correct response?
a) it’s inherently dangerous to expect drivers to yield, it’s much safer for people to judge the risks for themselves, and that goes for elderly and visually impaired people too. We should remove this unsafe facility immediately ; or
b) we must reduce the likelihood that drivers will fail to yield by reducing approach speeds, making the crossing more prominent and increasing enforcement.
When it comes down to it, doesn’t a) amount to simply placing undue responsibility on the weaker party? Which when you think about it is pretty close to victim blaming.
Maybe something of a duplication of the response I’ve made above, but I think it’s acceptable to not provide priority if there’s little or no delay as a consequence.
If delay is becoming a problem, then priority should be provided, and the measures you detail in b) should be employed.
“When a pedestrian is knocked down at a zebra crossing, which is the correct response?”
In practice, aren’t zebras being removed for exactly this reason? Certainly I am not aware of any new ones going in in my area, while there have been many new light-controlled crossings.
It is an enormous shame if this is the case.
Zebras are no less safe than signal-controlled crossings.
Cyclists are by far faster than pedestrians. A car driver can much more easily make a choice about what to do, and pedestrians, being much slower, can much more easily adjust their speed or direction in response to danger. Cyclists can’t stop easily. More easily than a car ever could, but not near instantaneously like pedestrians can.
There is a concept in the Netherlands, they are testing out what happens if you don’t use a circular cycle path when you give cyclists priority, but use the physical design of the non priority roundabout instead; does the better angles lead to sufficient safety that you can combine both cycle priority and safety? It’s not proven yet, but it does show some promise, one on a provincial expressway interchange at a turbo roundabout in Doordrecht and another one in Scheiden, West of Rotterdam.
Also, the Dutch don’t use roundabouts everywhere, just on junctions with 1750 cars per day per arm or less (I’m guessing that’s per direction) and a total of 25000 vehicles per day. And they often use other measures too, like grade separation when needed. Utrecht’s ring road in the west has been redesigned and one intersection provides a system where if there is too much wait and too much traffic with priority, a special signal gives the waiting traffic time to go, working part time, only when needed.
I agree with the analysis of the junction and approaches. The comparison in Assen, though, does seem to have much more space. The buildings at Perne Rd seem a lot tighter to the roundabout itself, and there may be limited land available to enlarge infrastructure. I wonder to what extent this influenced the design, whether it could have been improved within these parameters, and if so, why a better scheme did not emerge. Usually the answer seems to be: ‘there is abetter scheme, but it a) delays traffic too much; and b) it’s too different from previous schemes and would therefore fail safety tests, as it might surprise car drivers’!
I’ve added a size comparison at the end of the post – there seems to be around 10m *more* space available at Perne Road. You can get an impression of the scale of the roundabout from the size of the cars.
The roundabout island at Perne Road remains much larger, and I think that’s created some of the issues (although I understand it would have been expensive to adjust it).
Thanks! I am pleasantly surprised!
Priority only really becomes an issue with high volumes of usage. If the volume of (prioritised) motor traffic is low, waiting time is almost zero anyway. It works the same the other way too – where pedestrian footfall is low, zebra crossings get used and traffic waiting time is generally low. Where there is higher crossing usage there tends to be signalled crossings to ‘standardise’ waiting times (and these times can be easily changed to favour one or other mode). It would be possible to put toucan crossings in here that have an almost immediate push-button response to effectively give priority to pedestrians and cycles crossing.
An issue with designers not giving cyclists priority – and I agree that the question of cyclist priority or not is something of a side issue here – is that motorists will often yield priority to cyclists even when cyclist priority hasn’t been designed in. There are a lot of roundabouts in my local area where cyclists lose priority at each arm, but my experience has been that about 7 drivers out of ten will wait and wave me across anyway before they enter the roundabout, seeing me as “circulating traffic already on the roundabout” rather than as a “user of a cycle facility that hasn’t been given priority over the road at this location.” It’s very kind and courteous of them, but the ambiguity is an issue. Lots of polite nodding and waving and “After you!” – “No, after you, you have priority here!” does not make for a situation I would be happy to put schoolchildren in.
When I still lived in Galway and had to deal with multi-lane roundabouts on a daily basis, drivers trying to be nice were a very serious hazard. These days in Germany, with mostly single-lane roundabouts and cycle tracks around the outside, it’s more of a silly pantomime (although it can be tricky enough trying to gauge drivers intentions after dark and work out which set of imaginary rules they are working to – “cars have priority here” or “that cyclist is already on the roundabout and I’m waiting to enter it, so I should yield”.)
I think it’s easier on everybody to avoid that ambiguity by giving cyclists priority. I hold drivers up more by coming to a dead stop and then moving off from a standing start when they wave me across than I would hold them up by simply proceeding in the first place. They have to be ready to stop suddenly on the approach to roundabouts anway, expecting them to actually brake is no major imposition.
” is that motorists will often yield priority to cyclists even when cyclist priority hasn’t been designed in. ”
Agree with your analysis in general.
However one thing to bear in mind in Cambridge is the volume of cyclists. 58% of the adult population cycle, 1/3 of residents commute by bike. In this area there are two schools, which will have a very short but packed peak of parents and children cycling. It is one thing to give-way to one cyclist: another to give way to a dozen. Which is one of the problems of the roundabout: the waiting islands are limited if people start to stack up.
Please remember that your volume of cyclists is actually quite small by Dutch standards. What works in the Netherlands will work in the UK, even in Cambridge. Watch my video. You’ll see how children get home from school safely across a roundabout in a city which has more cyclists than Cambridge does.
The volume of cyclists is quite small, which means that whenever a junction gives priority to this “minority”, there is vocal opposition. I’ve seen this even in relatively cycle-friendly King’s Lynn, with letters in the local newspaper calling for one of the few toucans that has a possible sub-30-second wait for pedestrians and cyclists (NCN1 crossing A148 near Strikes bowling alley) to be made less responsive, even though that would only speed motor vehicles to queuing the complexes of road-road traffic lights either side of it! Mister Toad seems not to notice the vast numbers of traffic lights that let cars out, but one of the few that makes him wait for people to cross is a glaring aberration that sticks in the memory.
The volume of cyclists is perhaps ‘small’, but maybe still big enough for the cyclists to ‘stack up’ if they have to wait for gaps between the very large numbers of cars?
A video just before school starts would be informative.
Very good analysis Mark.
To the others with worries about priority I can only point out that it’s simply not nearly the issue that you think it is. Please watch just the first 80 seconds of this video to see how I use a roundabout between my home and the centre of Assen nearly every day. It’s necessary to slow down only in one direction and stopping is extremely rare.
Carry on watching the video past 80 seconds and you’ll get a flavour of the wide range of people who safely and efficiently use this roundabout every day. This includes disabled people, elderly people and thousands of children each day from local schools.
You’ll also see how the roundabout is used in conjunction with on-road cycle-lanes (it works very well, no drama at all) and how camber is used to slow motor vehicles, not just geometry as some try to claim.
Despite the heavy use, absolutely no cyclists, pedestrians or motorists were injured at this roundabout in the five years for which I have data and the four minor crashes which occurred involved only drivers shunting their cars against others due to their not concentrating on what the person in front was doing.
The Dutch are as impatient as any other people, and we’ve more cyclists than you’ll find anywhere in the UK, but no-one rides around the roundabouts here in Assen on the roads instead of the cycle-paths. You can invent an imaginary use such as that if you wanted to ride around and around the roundabout which would make it appear to be more efficient to be on the road, but that’s not a real life example of how anyone uses any roundabout.
For real life usage these roundabouts work extremely efficiently and extremely safely.
That video of Assen is one of the greatest pieces of roundabout design education that I’ve ever seen. There’s a roundabout near me and when in a hurry, I do ride on the road for part of it in one direction, but it’s not anything to do with priority: it’s because two of the arm crossings have obstructed visibility that mean I often have to push the button and wait the punitive minute’s delay for the toucan crossing to change, rather than ignoring the toucan lights entirely (as is legal for that type). The Highways Agency won’t cut the vegetation to improve visibility because there are toucan crossing signals… so I end up on the road, while more nervous cyclists would end up delayed and probably less likely to cycle at all because it takes too long.
Good roundabout cycleway design seems to have little to do with priority or traffic lights (which much UK discussion centres on) but everything to do with visibility and flow (which are rarely considered). The Perne Road mistakes include both poor flow due to the tight vehicle tracks of cycles turning to use the crossings, and poor visibility with its “over the shoulder” junctions where the designer apparently expects riders to look both ahead (for cycles coming from the crossing and to position themselves to cross) and over their shoulders (for approaching motors about to enter or exit the roundabout) as they approach a crossing. Not even motorists with all their mirrors are expected to look two opposite directions at once – why are they expecting riders to do it here?
As always thank you David, and I will take your comments and the vole’s on board. I can see that sight lines and vehicle speeds make a great deal of difference, but a steady stream of slow moving vehicles can be the most frustrating.
That roundabout in your video looks very quiet to me. Admittedly I’m not local to Cambridge so I don’t know the site in question, but there are lots of locations near me where no-one would ever get across a road if it weren’t for traffic lights, especially in the rush hour (and it’s usually a long wait). Plus there’s no way I want my kids trying to dash through a gap in traffic, and if you were trying to turn right (or left in the Netherlands) at a roundabout that’s a lot of crossings to wait at.
Tim: British observers tend to think the whole of the Netherlands “looks quiet”. Remember that having more journeys made by bike than by car will of course make roads less busy. Also remember that unravelling has deliberately taken motor vehicles away from cycling routes, leaving those places where cyclists go quite peaceful in comparison with the UK.
The roundabout also appears quiet largely because of the time of day when I made the video. I wanted to have children in the video so I went there at school chucking out time. However because children have the freedom to cycle in the Netherlands, this isn’t a busy time on the roads here like it is in the UK where so many children are collected from school by car. If you visit the same place at commuter rush hour then you do see more motor traffic, but many places in the Netherlands never become so snarled up with cars as is the case for apparently similar locations in Britain because successful policy has prevented this from happening.
Bear in mind that in the Netherlands roundabouts where cyclists cross on the level simply aren’t used on busy junctions. The maximum of vehicles allowed is quite low, and the places where they are actually installed tend to have even lower numbers. I don’t believe any Dutch planner would expect cyclists to be able to safely use any roundabout in the Perne Road location with the type and quantity of traffic which is typical on that ring-road. The closest equivalent roundabout sop far as traffic volumes are concerned that we have in Assen is shown in a different video. Note that even though traffic volumes are high there, at location there is no chance at all of a cyclist being hurt. This roundabout, like the other, has a perfect safety record (no cyclists, pedestrians or motor vehicle occupants have been injured).
We take people to both of those roundabouts on study tours and we usually see the one in the first video at rush hour. We also observe several other roundabouts for varying amounts of time, at different times of the day, and we visit roundabouts of the other design so that people can observe those as well. Some of these examples have less traffic, some have more and of course it all varies depending on what time of day we get to each location. We can do this and more in the study tour, but on the blog and in short videos it’s rather more difficult to show everything at every time of day.
Finally, please note that a left (right in the UK) in many cases actually means crossing only one arm of the roundabout. I do this all the time. You’ll see this in the first 80 seconds of the first video.
Just to reinforce what David Hembrow says: after cycling on the networks in the Netherlands and better German cities, I realised that junction and roundabout priority is not the big issue people think it is in the UK. Far more important for efficiency of cycling is the directness of routing, quality of surface, and continuity and width of cycle paths and tracks. Cycling in these places is vastly more efficient than in the UK because of the unobstructed character of the links. Giving way at roundabouts and crossings of major roads becomes a minor issue when the journey is mostly on direct, major links requiring few turnings. This is an argument for putting cycle infrastructure primarily along, or parallel to, main roads. Minor road cycle routes turn this around and turn a journey into the inefficient navigation of a maze. It’s how junction and roundabout details intersect with whole route design that is most important.
Priority matters when the level of motor traffic is so high that you will not find a gap otherwise. Even where we have high levels of cycling, we also have high levels of motor traffic. You can use traffic lights in those cases, but they bring their own problems.
Priority also matters where visibility is poor and things aren’t set back properly. The problem with losing priority over side-roads in the UK is that most of the time, the only way to safely cross is to come to a stop, regardless of whether there actually is anything there or not. Even low traffic levels lead to an interrupted journey, because you need to be ready for the time when there is something there.
Motors are usually backed up on the arms of roundabouts like this in the morning rush hour round me (suburban Surrey) and active school travellers simply walk or cycle across between the stationary cars. Given the comments from Cambridge locals, presumably the Massie roundabout situation is different, but as childbacktandem says above, a video of the school run would help an evaluation of this (PS there are some “after” videos at other times now on YouTube. The estimable Rad Wagon provides some for “before”.)
However, that’s toeing the LA party line too much for me, and as Mark T indicates, there’s no real excuse for not putting in cycle paths here to make the junction safe for everyone. Personally, I think the guiding principle shouldn’t be some notion of efficiency of cycling, or conflict mitigation, or traffic balance, but as David Hembrow pointed out on his blog just over a year ago, maximising childhood freedom.
It depends on the type of main road. If there are few turnings off of the main road, that’s a great place for a cycleway because it’s pretty hard to screw up with bad crossings of side roads.
I think this is one thing where the Netherlands being flat does help them, but not for the simplistic reason usually claimed: drains, canals and other waterways limit the numbers of road turnings across them. Railways and former railways are also often good for similar reasons.
And this is why I think the new London East-West cycleway along Embankment will probably work. No side roads turning into the Thames.
First of all it is rather uncommon for me to get into a public dispute about implementing “Dutch design” in a non-Dutch country. The Dutch design guidelines are based on the Dutch road conditions, the attitude of Dutch people and a set of traffic rules that might differ from other countries. The differences in the highwaycode might lead into different solutions to the same problem. Asked by several people to give my thoughts about this roundabout, I had a look on the initial post of Mark Treasure. But where to start with my response? There is so much to say and said already about the Dutch design principles and the selective use by the designers. The size of the green island in the middle? The absence of space between the pedestrian crossing and the roundabout itself? The absence of markings on the road to guide pedestrians? The absence of a clear devision between pedestrians and cyclists? The absence of a devision between cyclists and vehicles? The absence of back cover when taping in?
In my view the design isn’t clear for road users, not for car drivers, nor on a bicycle or as a pedestrian. The main question “where to put the cyclists” wasn’t answered by the design. The design team didn’t make the choice were to have people on bicycles (both arguments for having them separated or mixed can be defended, seen from the UK perspective), so people have to make the choices themselves. This is against the basic rule of predictability, one of the key elements of sustainable safety . Another rule that is violated is the need of forgivingness. It is violated by the lack of space between leaving the roundabout and the crossing, as outlined in the post of Mark Treasure. And I can continue why this design lacks the design principles for safe bicycle infrastructure. Although it might be an improvement of an existing situation, as the speed of car drivers has slowed down, the multi interpretable situation for cyclists is not an improvement at all. In the past it might have been clear that the roundabout was only accessible for brave cyclists, now it suggests it is a provision for all kind of people on bicycles.
I won’t go in the details of the reasons why this roundabout was designed as it was. The design team made a mistake, and that is not a shame. It would be a shame if they didn’t learn from this situation and persist in ignoring the basic design rules to create a safe environment for everyone. In the past we have welcomed politicians from Cambridge and Cambridgeshire, and also some planners and engineers for a taste of Dutch cycling. Now they start to implement facilities to make cycling available for everyone, we should encourage them to dig into the details of cycling and find out how things really work to make it safe for everyone.
Pingback: Turbogate: Bedford and beyond | The Alternative Department for Transport
The roundabouts with the non annular cycle path have several features that make them quite good. They usually have the cycle paths designed and routed in such a way that it avoids as many conflicts as possible, preferably on the lower volume arms. They have about 10 metres of a setback for cyclists when possible, 6 metres usual minimum. The angles are designed so that conflicts happen at 90 degrees when possible, and there are median refuges for the most part to let you break up crossings on bicycle or foot into fewer steps, designed so that you can fit a bicycle’s length within. The corner radii for cyclists is not too tight nor is it too wide, usually between 2 and 5 metres. The pathway widths are usually between 3 and 4 metres for a two way, between 2 and 2.5 for a one way, and the sidewalk in modern installations is usually 2 metres standard width, often wider. And as Mark Treasure explained, there is space between conflict points so that you can make more accurate predictions about who will be going where.
Here are things that come with Dutch roundabouts in general. A low design speed, 25-30 km/h, so any collision will be at a low speed, and more collisions are avoided. There is less flare out for drivers on the roadway, so you can see better and speeds are lower. The cycleway and footway crossings can be on a raised table, slowing speeds down as well. As David said, the camber helps to slow speeds. The circle itself is smaller. Cycleway crossings are usually crossing only a single lane at a time, and the roundabout is usually single lane. Bypasses around the roundabout are uncommon, but if there is a median refuge between roundabout and bypass for cyclists and pedestrians, then they also can be safer.
The volumes are usually lower. Between 500 and 1500 vehicles per hour on the busiest arm is best. Right turn bypasses could be sometimes added when the volume is a little bit too high and if there are cycle crossings on the bypass, the flow of the right turn is not so high that you are likely to stop to proceed.