Williamson Ave dragstrip. Auckland

The “flush median” is a pernicious road design that lingers in many places around Auckland and is still being utilised in many new road designs. I can only guess that its genesis originated in the late 60’s as a way to separate cars from the most severe of collisions, the head-on. It remains today, as that pesky give-way rule did, as a sort of monument to a particular era of traffic planning.

Besides providing a buffer distance between moving vehicles the flush median’s main purpose is to let cars turn left or right whenever they want, as infrequently as it may happen, while not obstructing the almighty flow of traffic. The flush median by design reduces the friction between vehicles moving in opposite directions and raises driving comfort and ultimately speed. Other such safety-minded designs such as the introduction of wide lanes, the removal of hazardous objects (aka trees), and over-scaled intersection geometry all are a form of “passive” design, an effort to physically design safety features into the environment. With the benefit of hindsight we know that this well-intended design philosophy combined with the desire for unimpeded flow often makes streets more unsafe, since it leads to speeding and driver inattention.

The unintended consequences of modern traffic engineering has now been widely documented on blogs like Strongtowns, and general audience books like Walkable City and Traffic, where Tom Vanderbilt sums up the situation nicely:

The pursuit of a kind of absolute safety, above all other considerations of what makes places good envrionments, has not only made those streets and cities less attractive, it has, in mnay cases, made them less safe.

So back to the flush median. There are opportunities in Auckland to recapture this space for better uses that in coordination with more progressive street designs will serve a wider range of users, notably people walking and on bikes. I’ve seen example of this wasted space along parks edges, in town centres, and even remnants lingering in the cbd– all places you would never need a turning lane, let alone a street design that encourages speeding.

A little trickier perhaps, but I see a street like Williamson Ave in Grey Lynn as a good candidate for such a rethink. The street is primarily residential in context, though it does serve a  morning peak hour pulse of traffic (including buses) in a sort of “grids gone wild” way.  Does it really need a flush median from end to end? I think it would be better served by a single central stripe with a few devoted (and short) turning pockets at a couple key intersections. This would open up the street to bike facilities which would have a significant network function of bringing people into the city centre.

Looking east along Williamson Ave towards Ponsonby Road, Auckland, 1963. (Sir George Grey Special Collections, Auckland Libraries)

There’s a lot that can be done in a street cross section when you “find” an extra 2.8m.  With changing values,  different economic circumstances, and a better understanding of how cities work, there’s no better time to reconsider the simple allocation of space in our urban environment. To return to a quote used before on this site from Enrique Penalosa:

“Urban transport is a political and not a technical issue. The technical aspects are very simple. The difficult decisions relate to who is going to benefit from the models adopted.” -Enrique Peñalosa

Residential through street, Melbourne. (Google Streetview)
Share this


  1. Those bike lanes in the Melbourne photo are door zone accidents waiting to happen. Better to put the bike lane to the left of the parked cars, then a physical barrier, then the parked cars (if there is room or need) and then the traffic lanes, a raised, but narrow, median that stops too many unexpected turns, then the mirror image on the right hand side. The bike lanes can be bidirectional and only on 1 side of the road. See brief description and photos – http://wellingtoncycleways.wordpress.com/2012/11/04/how-to-do-things-right-vancouver-style/.

    Proper bike lanes would have saved the Dunedin cyclist’s life last week. Spokes Dunedin is a better (albeit incredibly sad) blogpost than mine – http://spokesdunedin.wordpress.com/2012/11/10/editorial-comment-protected-bike-lanes-and-injury-risk/

    1. You’re right Matt. the cars should be against the lane and the cyclepath should be on the kerb side – raised preferably. Flush medians are the one piece of roading design that I hate. If a car is turning right then just stop, and wait. Safe as. Oh wait, that’s right, you wouldn’t want to hold up a car.

      1. But they’re no more of a door zone than the current situation with the added benefit that the cyclist is no longer sharing the same space as the traffic looming up behind. The presence of the painted bike lane is also an awareness tool for parking drivers. As an affordable next step this is an improvement.

    2. That Vancouver example is a totally different (cbd) context. Note the driveways at every house on Williamson Ave, these are problematic for “proper bike lanes”.

    3. That would also put cyclists in a dooring zone but remove the ability to also use the roadway… plus it would trap them between the footpath and the row of parked cars, leave them subject to cars pulling in and out of driveways blocking the cycle lane, and make them hard to see from the road due to the parked cars between them.

    4. Car doors open to about 1m from the body of the car. A basic cycle lane is 1.2m and it sounds like there is room on Williamson for 1-3-1.4m, leaving the outside of the lane is quite safe. I agree that it is not a perfect solution but it would definitely be an improvement over the current situation.

      1. A basic cycle lane is not 1.2m; that is an absolute minimum next to a kerb – 1.5m is recommended. If next to parking that goes up to 1.8m, with 1.6m as absolute minimum. So a 1.3-1.4m cycle lane here would not be an “improvement”; it would provide a false sense of security for an inexperienced cyclist trying to use it, invariably well within the ‘door zone’ (and passing motorists would wonder why some more experienced cyclists are moving out of the cycle lane into “their” space). Sub-standard cycle lanes are precisely what gives all of them a bad name (and causes problems like Dunedin), when good ones are an improvement over the status quo.

        1. Glen, what would you suggest with a 6.55m from centreline? Would anything simple be an improvement?

        2. It would be a squeeze and fairly marginal: you could have a 1.8m parking lane, a 1.75m cycle lane, and a 3.0m traffic lane to the centreline. But it’s the sort of width where I’d be more inclined to have a shared traffic lane with a few cycle “sharrows” (yes, not technically legal in NZ yet).

          Of course, if we remove one side of parking then life gets a lot simpler to fit things in (including the possibility of separated cycleways)…

        3. Cycle Action Network has this document that suggest minimum widths (not desirable widths) for retro-fitting:
          2 lanes, with parking both sides: 12.6m for wide kerbside lanes, 13.4m with cycle lanes and 15.9 with cycle lanes and a flush median.

          Drop parking from one side to get 10.3m, 12.2m and 14.7m, respectively.

        4. I know they are taking to ‘sharrows’ overseas but I have an issue with them – they are suitable for experienced cyclists. Without having used them but having read quite a bit about them, in my view they are not suitable for the young, old and inexperienced and it lets council off easy. “Hey, there you go – cycle infrastructure”

        5. I have used sharrows a lot when I lived in the US and to be honest they do nothing, and provide no help when cars are queued back which generally meant I had to switch to the footpath. With a cycle lane the cars generally queue out of the lane allow you to get to the front. There’s more than enough space to put cycle lanes down Williamson Ave and leave the parking. I’d agree getting ride of parking would be idea for high quality lanes but for a simple quick fix just paint the lanes down each side – common practice in places like Boston and NY and effective.

  2. Can I nominate Carlton Gore road for a future post. an amazingly wide road for minimal traffic yet the thin foot path, crowded with rubbish bins and temporary road signs, is heavily used by pedenstrians.

    A link between Newmarket and the Domain, it is a disgrace.

    1. Definitely. Carlton Gore Road is a shocker. Crazy wide road, and a footpath so narrow you can’t even walk two across along much of it, due to the rubbish bins (How did the buildings get planning permission without proper service lanes at the back?) There are more pedestrians than cars on CGR during much of the day. It could be a great walking / biking route and a fantastic social space with a bit of creative re-engineering.

  3. I don’t think the 1963 photo is what we should be aiming for, cars would drive 80kmh down a road that looked like that. That is just plain waaay too much tarmac.The flush medians are an improvement as a traffic calming device, as they do narrow the road space. However if there is no room for cyclists then are a bad idea.
    For example Carlton Gore Road would benefit from a flush median, however Cycle Lanes would be even better.
    Anyway don’t have to look a far as Melbourne for those painted cycle lanes, Palmerston North has had heaps added over last few years. Especially along College St such as here: http://goo.gl/maps/nG2Kz
    I agree they are far from perfect however they are very cheap and easy to add.

    1. I’m not suggesting 1963 is good- just demonstrating that streets change all the time. Well, at least they used to. “Paint is cheap” JS-K

      1. Yellow paint for no parking is even better than blue and green paint for bike lanes.

        But alas safety barriers made of paint aren’t as good as concrete.

    2. Why will cars cars would drive 80kmh down a road that looked like that?

      In Southern California the roads are much wider, and I didn’t observe much speeding.

      1. Generally, they won’t, but they will do 58km/h or so which is enough to make it hard to cross and if struck, most likely fatal for a pedestrian. Verbally I have received results of a traffic count / speed survey done by AT on a road that is very similar to that pictured above and 85% of traffic was between 50 and 60 km/h. Presumably some was even higher so there is a chance that just 8 or 9% of cars drive these kind of streets at below the posted 50km/h limit. Do we need to do something to the road design – you betcha!

  4. It would need a brave planner but what about removing parking from one side of the road, create a 30km/h parallel road on the other side with parking and a 3m cycle path between the 2 roads? That way drive ways are no longer a problem.

    1. Or, if it is a quieter street, use some of the kerbs for parking, create the 1/2m buffer and then paint the cycle lane.

  5. Even if the cycle lanes next to the cars is a door zone hazard (and it is) the advantage is that it generally means cars stay out it, meaning when traffic is backed up you can cycle to the front of the queue. Without the lane cars just lazily stop block the whole road. It would be a cheap thing to do, there’s really no reason for this huge median strip but you see them in many newly built roads. Take the bottom of Franklin Rd, it has the median strip, very wide lanes and not a cycle lane in sight – and this was just recently built and painted.

  6. Great post Kent. I’ve often wondered about these median strips and how much space is given over to them. To add a further suggestion to changing the way we paint; how about no paint i.e. no separation at all.

    1. For exclusively low volume residential streets I say yes. I’ve been photographing neighbourhood streets that have the most insane lane markings (a future post). Evidence shows that neighbourhood streets without lanes slows traffic since it introduces uncertainty for the driver. .

        1. I think this would only work on streets that are narrow already, and would have to be very careful when to apply it.
          I have seen many cases of very wide residential streets that have no painting, (centreline or otherwise) with cars travelling 60kmh to 70 kmh.
          Often the effect is that drivers feel they can drive faster as they are further away from parked cars, trees etc and they end up driving along the centre of the road.
          If it was done in conjunction with road narrowing and traffic calming such as parking delineation, adding cross-walks and other measures it could work, but would be sceptical seeing it up to 6000 AADT.

        2. Here is Hugo Johnston Drive. No centre line, narrowed road. 5 Day average was 3578 vehicles per day in 2009. Very nice to drive down. Imagine the amount of room you would have for a real cycle path if you narrowed one of those roads down to this? How much nicer for residents?


        3. That is a very unexpectedly well designed street for an otherwise dreary industrial area! Pity the land use is so inappropriate, a bunch of office park type places down a street that is impossible to serve by public transport! I do especially like the narrowing affect of indenting the parking and using different materials. This should be standard in residential areas.

        4. I know Luke. The first time I drove down there a few years back, I thought I had made a wrong turn. I can’t believe the traffic numbers. I know it’s busy but I hadn’t expected 3.5k. I’d love to see accident numbers but I suspect they would be very low as traffic does drive quite slow down that road. Proves a point though.

  7. a couple of months back and decided to “upgrade” the rest of the road by installing an over-the-top flush median along the entire length of the road. The result, of course, is that vehicle speeds have increased and the road is now more dangerous for pedestrians and cyclists.

  8. Marua Road in Ellerslie is a good/bad example of this. Following the road works to the Mount Wellington end of the street a couple of months back Auckland Transport decided to “upgrade” the rest of the road by installing an over-the-top flush median along the entire length of the road. The result, of course, is that vehicle speeds have increased and the road is now more dangerous for pedestrians and cyclists.

  9. They were revealing College Hill these last few weeks one side at a time, made me realise how absolutely massive this road is and yet has absolutely no provision for any road user except cars.

      1. Yes knew that, same applies to many large roads around Auckland. But since having the trams pulled out they’ve simply been handed over to cars. A blog post on road diets and complete streets which is becoming increasingly popular would be topical.

  10. I’m wary of throwing the baby out with the bathwater; flush medians can be a very useful tool. I can think of plenty of “traffic sewers” around Auckland that are four lanes kerb to kerb, when the volumes would suggest that a “road diet” of a lane each way, a flush median for turning traffic (and pedestrian islands) and bike lanes in the remaining space would improve things immensely. And, in contrast to the 1963 photo, I’m inclined to think that traffic lanes narrowed and confined by medians and cycle/parking lanes are likely to have a slight calming effect (that’s what we found when we investigated such a street in Chch – http://ir.canterbury.ac.nz/handle/10092/315).

    The elephant in the room is of course parking – do we really need to provide 100% kerbside parking in suburbia? (Oh the locals will grizzle initially, but they will adapt if you hold your nerve) Removing one side of parking and narrowing the median (except at side-roads) would allow for bike lanes each way. How does a 1m median help you (apart from reducing head-on crashes)? Simple, you still have the extra width of a bike lane if you need to get around a turning vehicle (letting any bikes go first of course…).

      1. “How does a 1m median help you (apart from reducing head-on crashes)?”

        There was an Auckland study (in the 1990s?) by Doug Wilson that found (though didn’t identify clearly why) that crash records on road with NARROW flush medians were better than on roads with wide flush median. Another positive impact of having side friction, I guess.

        PS: Carlton Gore Road was to GET a flush median based on the design plans as late as early 2014. We (CAA) split the dummy on this, especially as it would have meant that there wouldn’t have been a downhill cycle lane, just an unprotected uphill one.

  11. The Te Atatu Road widening is putting in a flush median. While I can see some of it’s benefits I do see it as wasted space. I did propose to Auckland Transport to use it as a bus lane in the peak time one way say between 7-9am and 4pm-6pm. I don’t know if it’s viable but would solve a problem. They said they don’t want to have a dedicated buslane out of the exisitng 2 lanes but I see Phil Twyford is querying this.

    1. Robert, I would like to see the 2 outer lanes actually cut off from the, shall we say 2 x traffic lanes, and use these as bus / T2 and residents. The interchange configuration is wrong as well as it keeps all the general traffic in the far left lane which is crap for residents. Roundabouts at Vera and Covill to allow residents to turn around to get in their driveways. Build a kerb to keep general traffic out. No flush median except at some road crossings – with lights. 37,000 cars per day (5 day average). Unless they give people a good reason to use PT (bus lane or train) there can be no good outcome for Te Atatu South and Te Atatu Road.

      1. I like how you’re thinking out of the box with the roundabouts. 4 sets of traffic lights as they propose is not ideal. What they propose will still see buses caught in the traffic so it would be good to see a T2 lane and change the interchange layout. I’d prefer the entrance to the suburb not to turn into another Lincoln Rd.

        The flush median I guess was also proposed for people crossing the road with 4 schools are in the area. I guess safety is a priority for pedestrians.

        1. I’ve thought of pedestrians as well. In my plan there would be a couple of additional traffic lights – but not for cars, just pedestrians and cyclists. Short phase as per the Netherlands. This would be much better than what is there now. Also, something needs to be done about ‘rat running’ through Vodanovich Road and the surrounding streets.

  12. It is there also to serve as a traffic calming measure, reducing the lane width while still obtaining access to properties and roads, not the most effective however. In my opinion a better traffic calming technique to bring us into the 21st century would be dedicated cycle lanes, killing two birds with one stone!

  13. Well I’m going to stick my neck out here and say that I disagree with Kent’s dislike for painted median strips. As a pedestrian I’m very fond of them because it breaks the task of crossing a road into two stages and makes it much much easier to get across a road. You don’t find yourself stuck in the middle turned sideways trying to squeeze onto the single line separating the two flows of traffic.

    1. Personally I feel just as unsafe in a median strip as on a line, mainly because I keep having to watch out for drivers that might decide to drive down the median strip with no warning. Changing driver behaviour would be better. Crossing a road like this http://goo.gl/maps/ZPMrW is easy when the drivers notice you waiting, see that there’s a gap in the oncoming traffic, and slow down to let you cross. Would never happen in NZ I bet!

    2. @ Joshua Arbury: Median strips per-se are not the issue. It is the “painted-on” variety that risks creating a race-track environment. A better alternative is the kerbed or raised-median which cannot be driven onto at speed and therefore cannot be claimed as de-facto extra road-space to encourage the antics of speeding drivers. It also provides a more secure refuge for crossing pedestrians and can be planted with bushes or installed with traffic-calming features to counter the “race-tracK” environment. Best of all may be where it is interspersed with stretches without any median at all, thereby creating a “wavy” effect in the road alignment and discouraging lead-footing. However those stretches where the median is present must not restrict the lane-width so much that cyclists are forced into conflict with cars. This problem definitely occurs at certain kerbed “traffic islands” installed on narrower roads for traffic calming purposes, at which the resultant lane-width compromises the separation between large vehicles and cyclists.

  14. What about 4-way stops, kerb bulb outs and crosswalks at regular intervals? Not so good for flow, but good for local movements. I think the median and pedestrian refuge is a second class ped facility.

    1. I’m mainly thinking of arterial roads here I suppose, which perhaps is a bigger question about Williamson Ave and whehther it’s an arterial road or not. Just I know that crossing say Jervois Road is easier than crossing Onewa Road, even though Jervois is 4 lanes and Onewa is (generally) just two lanes. Because you can bite it off one chunk at a time.

      1. Williamson is treated as an arterial but it shouldn’t there’s Gt Nth Rd right next door which has been overloaded with traffic, Williamson is used as a rat race to avoid it.

        1. I’ve never seen Gt Nth Rd overloaded with traffic ever, not in the stretch where it runs parallel to Williamson. Perhaps I lead a sheltered life.

  15. I’m not sure about this one Kent. As others have noted the flush median in question is not really an alternative to cycle lanes – because the latter would take up more space than the median.

    The flush median also tends to benefit pedestrians in two ways – 1) it narrows the lane width so that cars tend to travel in a more predictable line and 2) it provides opportunities to cross halfway.

    What about turning it into a hard raised median with a two-way cycle lane down the middle? Or if that’s not feasible just a hard median with trees etc?

    1. There’s plenty of space in that median for cycle lanes down each side – using it as a two way cycle lane in the middle did cross my mind and with cycle numbers so low in Auckland it would probably work as there would rarely be people passing in opposite directions.

    2. Now I’m going to disagree with you Stu. I’m no fan of hard medians as I really do think that they create a “motorway” environment and militate against it being easy for pedestrians to cross the road.

      1. Depends on the design I guess. Some are planted up in a way that makes them difficult to cross. Others have a step that would make it difficult to use with a push-chair etc. Plus I think they create more of a motorway environment by reducing friction and limiting turning movements.

        1. Hard medians have one major issue in being a barrier to emergency vehicles, which all too often are ignored. Hard medians should be avoided on any routes that are not dual-carriageway, the last thing you need is an ambulance having to bump its way over a hard median or to be blocked because traffic has nowhere to go to move out of its way. I’m strongly in favour of painted medians for the reason Josh said – pedestrian refuges, and although clearly it doesn’t matter to most on here, reducing the incidence of head on collisions actually matters.

        2. The results of a 2 car head on accident at a combined speed of 100 km/h (we are discussing 50km/h areas here) are, these days, not usually fatal. The risk to a pedestrian at 50km/h is far in excess of that faced by the car occupants.

    3. Two way cycleways in the median are downright deadly. I tried one in my old commute in Australia, gave up after one shot and went back to riding in the regular lane.

      Problem is all cross and right turning traffic will cut across your path. In the case of right turning traffic they are coming from back over your left shoulder or far off to your right. The sum result is you have to come to a dead stop every half block or so, make a full 360 degree sweep looking for conflicting traffic, ride 50m more to the next cross point, come to a dead stop…. etc

    4. Hi Stu

      Often the question is not “will the flush median removal give us exactly enough space for cycle lanes”, and more “will the flush median removal give of 80% what we need and can we get the rest by some slight tweaks to lane widths and kerb locations.

      Without removing the flush, it’s often pretty much impossible.

  16. The above street is pretty close to having space for bike lanes. (5 cm) http://nacto.org/cities-for-cycling/design-guide/bike-lanes/conventional-bike-lanes/#design
    Whether lanes here are the best solution I don’t know. I do think that there is little parking ‘turnover’ so potential for dooring hazard is less than say Ponsonby. I’ll need to keep working on you guys. The reason medians are good for pedestrians is since there are few, if any, proper street crossings. I’d be interested in how many people would let their 8-year olds cross this street. Sidenote: Why are mothers with prams waiting in the middle of Posonby Rd on little islands?

    1. But these medians only become useful because as you say there are no crossing – coming from Zurich it is a shock in Auckland because of how few crossing opportunities there are. A street like Williamsom Ave probably needs at least 10 along its length – not only would these return it to its use as a local street, it would reduce barriers to pedestrian movement, and allow cycle lanes to be installed. Even arterial routes outside of cities in Switzerland signposted at 60+ would have regular crossings par de cours.

  17. I moved here from the US five months ago, and I find the “flush medians” here to be extremely odd. First, they’re often actually too narrow for a car to fit, so they don’t help traffic at all. In fact, they confuse people. Second, in other countries has marks like that mean DON’T GO HERE so I had to learn that you actually CAN use these lanes.

  18. From a 2008 study Provision for Cyclists on
    Williamson Avenue, Grey Lynn “…hourly speed data for each day reveals periods when 70 percent of vehicles exceed the speed limit at 77 Williamson Avenue. The majority of speeding is in the 50 to 60 km/h range. The percentage of vehicles travelling between 60 and 70 km/h is approximately 5 percent on weekdays, but jumps to 15 percent in the 7 to 8am period during the weekend.
    For 101 Williamson Avenue there are daytime periods when 45 to 50 percent of vehicles exceed the speed limit. The number of vehicles travelling between 60 and 70 km/h is 2 percent on weekdays, but jumps to 10 percent in the 7 to 8am period during the weekend” & “While the daily totals are higher on Great North Road, the morning peak hour figure for Williamson Avenue is higher than for Great North Road. The afternoon/evening figures are also similarly high

  19. The other thing that occurred to me is that while narrowing a road and building kerbed parking bays and a separated cycle path is expensive in the first place, think of the amount of resealing work that no longer has to be done in the future? Probably 50% less at a guess.

Leave a Reply

Your email address will not be published. Required fields are marked *