The 85th percentile as a tool for improving roads and streets

The “85th percentile” speed is a speed at which 85% of traffic will be travelling at, or below, along a street or road (under free flow conditions). It’s typically associated with the setting of speed limits, and (more controversially) often used as an argument against lowering them, or enforcing limits.

In particular, some police forces have been reluctant to enforce 20mph limits that have been introduced on roads that previously had a higher speed limit, without any changes to the design of the road, on the basis that enforcing this lower speed limit will prove to be too much of a drain on their resources – too high a proportion of drivers will be exceeding the new (lower) limit.

I have to admit I have changed my position on this issue over the last few years. Previously, I had been of the opinion that a speed limit is a speed limit, and that it should be enforced, regardless of how many people are breaking it. That any refusal to do so was effectively a ‘cop out’ (excuse the pun) on the part of the police.

But I think the police (or ACPO) are exactly right when they say

Successful 20 mph zones and 20 mph speed limits are generally self‐enforcing, i.e. the existing conditions of the road together with measures such as traffic calming or signing, publicity and information as part of the scheme, lead to a mean traffic speed compliant with the speed limit.

To achieve compliance there should be no expectation on the police to provide additional enforcement beyond their routine activity, unless this has been explicitly agreed.

In other words, the 85th percentile speed (the speed at which 85% of drivers are travelling at, along a road) should correspond much more closely with the posted speed limit through the kinds of measures the police list here – in particular, the design of the road. Research carried out for Manual for Streets shows that the speed at which drivers travel along a road is influenced by its design – principally its width, and forward visibility. If plenty of people are breaking a limit, that probably tells you either the limit is wrong, or the design of the street is wrong. Something has to give.

And this is the reason I am suggesting that the ’85th percentile’ could actually be a force for good – it cuts both ways. While it can be used to reinforce the status quo, it can also tell us that the design of a road is inappropriate for the posted speed limit.

Take, for instance, a situation in which a residential street with a 30mph limit has that limit lowered to a 20mph limit, without any changes to the design of the street, or to the motor traffic network. Let’s then say that the 85th percentile speed of motor traffic on this street, after the introduction of the lower limit, is much more than 20mph – close to 30mph, for instance.

What does this tell us? It tells us that the design of the street isn’t doing its job. While it might be a good idea in the short term to get the police out with speed cameras, a long-term solution should be to change the nature, character (and usage) of the road so that the 85th percentile speed on it is much closer to 20mph.

So the 85th percentile is an effective way of demonstrating when speed limits and road design are out of kilter. Take, for instance, this 20mph limit on Midland Road in London, running between St Pancras and the British Library – just one of many main roads in London that have, in recent years, had their limits lowered from 30 to 20mph without any change to the design of the road.

Screen Shot 2015-08-20 at 13.35.03

I don’t know what the (free flow) 85th percentile speed of motor traffic is here, but I’d be willing to bet good money it is way, way over 20 mph – this is a wide road, with three lanes of motor traffic bearing down on Euston Road, in one direction.

Again, we could get the police out here with speed cameras, but really, the discrepancy between the posted limit and the way people are actually behaving on the road tells us that something more serious is wrong here – the messages the road is sending out to drivers don’t correspond to the limit that has been painted on it. Something has to give.

By contrast, in this early-1990s 20mph zone in Horsham – designed to be self-enforcing – it’s pretty much impossible to drive at 20mph (despite it being one-way!).

Screen Shot 2015-08-21 at 10.01.18

A combination of speed humps, tight corners, limited forward visibility and surfacing means that the 85th percentile speed is likely to be at (or even below) 20mph, which tells us that the speed limit and the design of the road are in agreement, and there’s little or no need for enforcement.

The same logic can be applied to 30mph roads too. This road in Wageningen, NL, has a 50km/h speed limit – and it’s reasonable to assume that the 85th percentile speed will be at or below that speed, due to the design of the road.

Screen Shot 2015-08-21 at 10.04.40

The carriageway is very narrow, with motor vehicles barely passing each other, and has no centre line.

And there are other ways of bringing the 85th percentile speed into line with a 50km/h (or 30mph) limit on these kinds of distributor roads – for instance, pinch points for motor traffic (that don’t affect cycling).

Screen Shot 2015-08-21 at 10.08.57

So if the 85th percentile speed on a 30mph road near you is (under free flow conditions) closer to 40mph, that should tell us that action is needed to bring driver behaviour more closely into line with the posted limit, through these kinds of measures. Principally, perhaps, by reclaiming a good deal of the carriageway for cycling, consequently narrowing it down for motor traffic.

I hope this explains why I’ve changed my mind, and why the 85th percentile can be a constructive tool for improving streets for walking and cycling!

This entry was posted in Uncategorized. Bookmark the permalink.

42 Responses to The 85th percentile as a tool for improving roads and streets

  1. A point which you didn’t mention is that, while local council officers will not reduce the posted speed limit on roads where current speeds are significantly above the proposed limit, they also say that reducing speed limits is ineffective because it does not significantly reduce actual speeds. Of course it won’t if the only changes that are made are on roads where the speed was already low. It seems to me that the underlying issue here is an attempt avoid any reduction in speed limits in order to prioritise vehicle throughput and to avoid the expenditure and possible public criticism from making changes.
    A second point is that I’m not sure how an 85th centile is calculated and whether it is necessarily the most useful measure. For example a wide, straight road may have traffic moving at under 10mph in the rush hour, but have Friday night yobs travelling at 50mph. Perhaps a highways engineer could comment.

    • Andy R says:

      “A second point is that I’m not sure how an 85th centile is calculated and whether it is necessarily the most useful measure. For example a wide, straight road may have traffic moving at under 10mph in the rush hour, but have Friday night yobs travelling at 50mph”.

      TA 22/81 explains how vehicle speeds are measured and analysed. Essentially it’s done in free-flow conditions, which effectively rules out the morning and evening peak hours (unless you specifically want to compare peak hour speeds to the rest of the day).

      Click to access ta2281.pdf

    • Alex Hosking says:

      I think some councils have started to do exactly that, not setting speed limits below the 85th percentile, but far below even the mean average speed on the road with no changes to the engineering of the road.

  2. Notak says:

    Yee-e-es… But this seems to be viewing 20mph, and speed limits generally, as primarily a safety measure. I’d say they should be set as much by considerations of environmental (noise, pollution, wildlife) and comfort (can you cross the street? hold a conversation? etc) factors.

    There’s no doubt the increase in road widths and smoothness over the last hundred and more years has allowed traffic speeds to increase and in many/most cases been done to enable that increase, but equally, speeds have increased on urban streets which are now narrower (due to parking and loading) than they were before the motor era. True, that’s now on tarmac rather than cobbles, setts, wood or mud, but it does indicate that the road layout is not the only factor in determining speeds (besides, do we want to cycle or even walk on those surfaces?). In addition, there are many locations where enforcement has resulted in lower speeds on a long-term basis, ie after the enforcement has ceased. Design is important, but achieving decent speeds (in any sense of decent) should not be its only, maybe not even its primary, goal; nor is it the only or in many cases the primary influence on drivers’ speed.

  3. ORiordan says:

    Apologies for the long post but this is a very apt topic for some investigations I’ve been doing in my street in West London and I’d be interested in other people’s experiences.

    My street is totally residential but used as a rat run by drivers, particularly in one direction. Residents were concerned about vehicle speeds so the borough installed a 20mph limit on the street with some signs that flash 20mph if a vehicle is above a certain threshold. I don’t know the measurements before the 20mph limit was put in place, or the speed threshold for the flashing sign.

    The perception from our residents group was this made no difference whatsoever to vehicle speeds. I put in a FOI request to the council asking for speed measurements from the sign and they replied saying they had no way of getting this.

    I then went DIY and have been using software called Raser Abwehr (German for “speed defence”) with a webcam pointed at the street and a measured distance marked on the road. The software works out the speed of vehicles between the measured distance, captures images of the vehicles and records the data so it can be processed in a spreadsheet.

    I excluded two wheelers and vehicles travelling below 10mph (of which there weren’t a significant number) and after crunching the data for a week’s worth of measurements between 9am to 7pm, the 85th percentile speed was almost exactly 30mph.

    Just as the residents had suspected, drivers are ignoring the 20mph signs and treating the street as a 30mph road. I’m sure this is because it *looks* like a 30mph road as it is dead straight for about 250m.

    We’ve forwarded the data to the council traffic department and local councillor, however there is a borough-wide 20mph consultation at the moment, so I think nothing will be done until this is finished. I think the best approach will be to raise the issue during the 20mph implementation, assuming the overall consultation is supportive.

    I think the tough thing here will be to get the council to propose some solutions. Being cynical, I’m anticipating responses of, “well wait until the borough-wide limit has been in place for a while”, or “we’ve installed signs, nothing more we can do…” In other words, 20mph is really a “tick the box” for road safety rather than actually changing street design to achieve it.

    • Notak says:

      Just as the residents had suspected, drivers are ignoring the 20mph signs and treating the street as a 30mph road. I’m sure this is because it *looks* like a 30mph road as it is dead straight for about 250m.
      Well, the reason it looks like a 30mph road is also because the vast majority of such streets do have a 30mph limit. In some countries until recently (France, various parts of Eastern Europe, for instance) it would ‘look’ like a 35mph road, because those countries had a 60km/h urban limit. So to an extent the attitude that when 20mph is the norm and has been for some time, speeds on these streets will come down, is correct. Trouble is, obviously, that’s a long time coming.

      • ORiordan says:

        Yup, I’m fully expecting that the council’s response will be to wait until a borough-wide 20mph limit has had enough time to change behaviour. It will be interesting if they can be pinned down as to what that time may be.

    • Dan K says:

      Making the street access only is the cheapest solution. Creating chicanes and build outs is expensive, reduces parking so is unpopular and is only partly effective. Filter, filter, filter!

  4. The problem with the 85 percentile test is it encourages authorities to think in terms of individual roads, rather than on an area basis.
    The reason Cambridge switched to an area-based approach was that individual roads across the city were 20mph where there was resident demand, resulting in a constant changing of limits within the city. There were lower limits and traffic calming where there were engaged residents, and nothing for others. A TRO was required for every street, which meant lots of individual costs and officer time on limited projects.
    A project to traffic calm every single street would have been prohibitive, but one to just change the limits was manageable, and more efficient than treating one-off requests that came up. Now every residential street is 20mph, (apart from for those unlucky enough to live on the A roads, which were excluded, and those B road which were consulted on). There are still problems, but now it’s much easier to say “the limit is 20mph but people aren’t sticking to it, so now we can look at traffic calming”. Harder to justify such big measures where people are doing 35 in a 30, even if the true purpose of a particular road should be residential rather than transport corridor.
    And the last point is really the key one. It’s moving towards a principle that urban streets are for people, not for moving cars. Limits aren’t a panacea, but are a part of the argument for the purpose of streets.

    • Kie says:

      The beef I have with traffic calming measures is that they are horrible at any speed if you are on a road bike, they need to have less abrupt start/finish or have a gap for cyclists.

      • Kie says:

        Ps, referring to road humps and the like, chicanes are good.

      • Ben Hicks says:

        Cyclists are often a huge part of the problem and in themselves should be slowed just like their four-wheeled counterparts. Cyclists moving at 30mph along roads set at 20mph create a whole different dynamic for pedestrians and car drivers.

  5. fred says:

    The problem here, of course, is that streetworks are expensive, whereas putting signs up is cheap.

    It would be useful to know how effective average speed cameras can be in these locations. They’re certainly very good at ensuring compliance on motorways..

    • ORiordan says:

      I think the issue for average speed cameras will be their cost for the number of vehicles measured.

      A residential street may have a couple of thousand vehicles a day while a major motorway may have a hundred thousand vehicles a day.

    • Kie says:

      I think the best solution is to make most residential streets into dead ends with cut-throughs for cyclists and pedestrians.

  6. T.Foxglove says:

    Our Local Authority & Police use the 85th percentile the other way. So outside of town/village centres, if the road has 30mph limit & traffic are doing 40mph and you complain about speeding, then the speed limit is lifted to 40. Problem solved.

    Changing a sign is cheaper than engineering or enforcement.

    They also refuse to drop the speed limit, so you get things like this:

    This is a 40mph road & that is scheme funded by the cycle safety fund. It was designed so vehicles had to drive in the cycle lane as they were concerned about head on collisions between vehicles at 40mph.

    It was suggested that if they dropped the speed limit then lane widths could be reduced to allow a mandatory bit of paint. 85th percentile was used as a threat that the limit should be raised to 60mph due to the rural nature of the road and a lack of accidents.

    • Herb says:

      That’s how the 85th percentile is typically used in North America as well. It cuts both ways: as a tool it can be used to either justify adding traffic calming measures, or just increasing the speed limit.

      One big drawback of the 85th percentile is that it excludes the speed of cyclists and it is the only measure that traditional transportation departments use to gauge road “safety”. The theory is that the most dangerous part is not the *average* speed but the speed differential. So if the planners are ignoring cyclists they are also ignoring the most obvious speed differential.

      In Toronto, where I live, the 85th percentile been used both ways. The transportation department has made it quite difficult for communities to reduce the speed of vehicles on their street. Resources for installing traffic calming measures are limited and a community is required to jump through hoops before anything is installed. The 85th percentile is used to justify that the status quo is fine. Only recently has the downtown community council (representing municipal politicians downtown) been able to reduce the speed limit on all residential streets. By making a blanket reduction, the communities can then request traffic calming measures from the transportation department by using the 85th percentile.

    • Bobo says:

      It works because then pedestrians will be less complacent, and know to be more careful.

    • Mark Williams says:

      The x-height of text on road signs is also supposedly based on 85th %-ile speeders—even on brand new roads (no, me neither). Yet if you go around measuring it on existing signs and looking up those sizes in draft chapter 2 of TSM (or even appendix A of LTN 1/94), it is clear that some are significantly larger than necessary for 85th %-ile of current speeders, design speed of road or the present speed limit. Whenever the issue of speeding is raised, the council throw their hands up in mock surprise that anyone would possibly speed on `their’ roads whilst themselves erecting signs which can be comfortably read at double or triple the speed limit.

      At the same time, you have the DFT simultaneously complaining about one type of signage `clutter’ and then hypocritically turning a blind eye to the much more oppressive, costly and visually intrusive `clutter’ of signs excessively larger than those [already over-generously] prescribed; which surely serves to encourage speeding. Mysteriously, the attitude of TPTB when it comes to cycling-specific signage is exactly the opposite—with London city council at the vanguard of strongly discouraging signs which are legible to even stationary cyclists equipped with binoculars…

      Of course, if the police force cannot be bothered to enforce the speeding law—or, indeed, most other road traffic law—then perhaps it is time to start basing their over-staffing levels on 85th %-ile of willingness to actually do their job?

  7. Jim says:

    Some traffic calming methods have the (presumably unintended) side-effect of slowing cyclists down too. That definitely has to be borne in mind when they’re being installed.

  8. gregoryiain says:

    I think this is an important piece. Riding in France this year in a wide range of environments and landscapes the road-design and not signage or driver-choice was the greatest contributory factor limiting speed, choice and (it seemed to me) stress upon all road-users making my time cycling there unrecognisable from cycling in the British Isles.

  9. Ian says:

    Shame that drivers can’t simply be expected to obey the law. Perhaps proper punishment of offeders would be a cheaper alternative, tjough some mix of design and serious enforcement might be best.

  10. Andy R says:

    The shame of it is that most Local Highway Authorities have at the very start of their Design Guides a heirarchy of roads. Picking the Herefordshire Development Design Guide (1) at random, it has; Strategic Routes, Main Distributors, Secondary Distributors, Link Roads, Local Access Roads and Rural Access Lanes. Whilst this is more categories than some other LHAs use it still provides a framework upon which Sustainable Safety (2), both in terms of ‘Functionality of roads’ (monofunctionality of roads as either through roads, distributor roads, or access) and in ‘Predictability of road course and road user behaviour by a recognizable road design’ (road environment and road user behaviour that support road user expectations through consistency and continuity of road design) could be implemented. The problem being we don’t stick rigidly to those definitions, and muddy the differences between roads serving different functions both in their physical layouts and the speed limits chosen.


    • Mark Williams says:

      Outside the public sector, allowing a random/ self-selected bunch of amateurs to choose the speed limit after the `professionals’ have laid out the road would not inspire much confidence in the competence of those `professionals’. Then they have the nerve to hype themselves `engineers’ and moan when taxpayers do no show them the reverence they believe is due to them.

      These same `professionals’ often proudly boast about every metre of `their’ highways being hand-bodged from scratch as though this were a good thing in and of itself. Presumably this is intended as a slur against those countries which `merely’ cut-and-paste pre-approved highway designs from a menu on the grounds that it’s the only proven way of getting reasonable consistency and cost control. But it actually just makes them look like Luddites who have too much time on their hands and could do with a few rounds of thinning-out.

      And don’t get me started on the `consultants’ ;-)…

  11. Jitensha Oni says:

    There’s driving style as well – I guess that is what gregoryiain refers to. Examples.

    I still have the occasional chuckle at the time I was on my bike following some car in a 20 mph street in Strawberry Hill where those lopped off pyramid-stylee road bumps were used to effect speed reduction. The driver bottomed out the car on the tarmac over every single one: they never did get the timing of the accleration-braking right. I guess seeing the bike rider not diminishing in size in their rmirror didn’t help. OK, that’s from above the 99th percentile, but Maple Road in Surbiton is a straight 20 mph road with schools, shops and cafes and has some well constructed speed bumps that extend right across the carriageway. Many have zebra crossings on top. They don’t jar when cycling and neither do they if you set your speed at a constant 20 mph while driving. This doesn’t stop most drivers accelerating between the bumps, braking, accelerating, braking etc. That’s most drivers – from the 15th percentile? Is this a safe way to drive, even if they’re averaging 20-25 mph? I don’t think so. Would other interventions do better? Not sure. On the other hand I’ve had a number of non-cycling friends from out of the area remark how insane the driving is in SE England, so maybe other areas would be better primed. Still, any intervention should be able to handle the diffcult cases, wherever they occur.

    Otherwise, CamV’enne makes an excellent point about planning over blocks of streets. It’s still the bad old days in mini-Holland Kingston-upon-Thames. Here’s a 30 mph residential cul de sac- yep, cul de sac – off a 20 mph minor distributor.

    • RobertL says:

      Here’s some good traffic calming:,153.0416297,3a,75y,226.43h,62.44t/data=!3m6!1e1!3m4!1sNy1q701k0owFYiOFNT3dPw!2e0!7i13312!8i6656?hl=en

      This is Arthur Street, in New Farm – about 2km from the CBD here in Brisbane, Australia. The street allows parking on both sides and has “bike awareness” stencils along each side – in the door zone – so it’s not perfect.

      However, it also has these wide table-top speed bumps. The speed bumps do not extend across the entire street, The green painted gaps at either side are for cyclists to ride through without having to slow down, and the little traffic islands are to prevent parking at those points so that cyclist have the necessary room.

      This shot is looking uphill, but the design means that when I ride home downhill, I can coast at ~40km/h (it’s a 50km/h zone) and overtake cars that have to slow for the speed bumps. That never fails to surprise them : – )

  12. John C says:

    I think that some drivers possibly need some training as to how to drive in 20 mph zones. I have heard some drivers say that their car won’t go that slow, bizarre I know.

    However if driving in 20 mph zones was part of the driving test they would know that they don’t have to go everywhere in 4th or 5th gear.

    Changes to the driving landscape such as 20 mph zones could be a good reason for introducing re-testing at intervals in a drivers career to ensure they are up to date with such changes. Including things that are covered in the Think! campaigns such as to expect people riding bikes to be more than a door width from parked cars and that on narrow lanes they can expect to see bikes in the middle until it is safe for a motor to pass.


    • Ian says:

      If the licence and car were taken from them for offending they’d soon discover how to drive below 20mph.

    • Har Davids says:

      A car that won’t go as slow as 20 mph seems like a technical anomaly to me. As far as I know they often do 0 mph (about 90% of the time).

  13. Harry says:

    The exciting thing about dropping the limit is that the 85th percentile can then be posed as a problem. For example, the Royal College Street upgrade proposal noted that the 85th percentile speed was around 30mph (I forget the exact figure). Given the speed limit was now 20mph, this was a problem. While the speed limit was still 30mph, it wasn’t. So, to my mind, the more we can get speed limits lowered, the more we can then point out that it hasn’t worked and more action is needed…

    • Herb says:

      That is the argument used by our downtown Toronto politicians who recently dropped the residential street limit to 30kph. Perks, one of the politicians said that now the 85th percentile will now show that traffic calming is required.

      The Transportation department, however, has typically opposed dropping speed limits because there’s no traffic calming in place, and has also opposed putting in traffic calming first because the 85th percentile (with the old speed limit) shows that speeding isn’t a problem.

      A catch 22 that allows them to do nothing. So looks like the solution was political to force it open.

  14. The volume plays a larger factor though. Dutch cyclists in the rural area are OK with mixing with 60 km/h traffic, but only in low volumes. Keeping the speed low is a worthy goal though. 4.5 metre wide carriageways (between the two kerbs on either side of the 4.5 metres, this could either be the curb between the footway and the carriageway or between the kerb that parking vehicles must cross) with brick paving, unmarked, unsigned, raised junctions with very tight corner radii, preferably with a bend in the road every 75-100 metres or so, and the inability to use the road for any reason other that you need to use that particular road to get from your destination to the distributor road, helps.

    This also makes me think about motorway speed limits. Why doesn’t the UK metricate and mostly use 120 or 130 km/h speed limits like most of the rest of the world on their motorways? The traffic is mostly going that speed anyway and crash risks are lowest when traffic moves at the same speed, regardless of exactly what that speed is.

    • Herb says:

      It’s true that traffic volume makes a big difference.

      But it isn’t a typical Dutch default planning decision to have cyclists mix with car traffic going 60 km/h. The Netherlands has a very extensive path network throughout the country so that people only rarely need to bike on roads and even then only on roads with low traffic and low speeds.

      On my last trip to visit relatives in the Netherlands, my wife and I bought the thick guidebook of the path network and used it extensively in South and North Holland. We only briefly had to mix with car traffic and I recall it only happening in towns, not in the countryside. We felt our stress drop–it was very calming biking there.

  15. Pingback: Kingston Go-Cycle summer proposals, Part One – Stuff Rich Writes

  16. Pingback: Asking the wrong questions | As Easy As Riding A Bike

  17. Pingback: Designing for Failure: Wheatfield Way – Stuff Rich Writes

  18. Alex Hosking says:

    At last another cyclist who understands how speed limits work. A 40 limit road near my has a compliance rate of 97% and an average speed of 31mph, another road not far away with a 30mph speed limit has a mean average speed of 42mph, that’s the level of disconnect we’re dealing with.
    Often it’s the way speed limits are set that are in of themselves the thing that exacerbates people’s contempt with speed limits. Like for example having the speed change before the road does X distance from where it needs to, people get used to speeding limits changing seemingly for no reason, it teaches bad behaviour.
    Normally the response when trying to explain why you end up with such ridiculous levels of non-compliance is to just chest beat and shout louder “Yeah, but people should just obey speed limits”.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.