Why the green man goes off before you get across.
Every traffic light has a little computer in it. When you press the button, it decides how many seconds you get to cross.
It does one sum.how wide the road is, divided by how fast you walk7 ÷ 1.2 = 6 seconds
It thinks everyone walks 1.2 metres a second. That is a brisk walk.Somebody measured people crossing a road in the 1950s and wrote that number down. It has not changed since.
Most people over 65 walk slower than that.Three in four men. Five in six women. Measured at home, in England, by nurses with a stopwatch.
So here are two people at a crossing. One walks as fast as the computer thinks. One walks as fast as most women over 65 do.Scroll down and press the button.
One road, 7 metres. The dot on the left walks at 1.2, what the light assumes. The dot on the right walks at 0.8, an average woman over 65. They both reach the kerb as the green man is ending.
The green man does not mean "you have this long to cross". It means "you can step off the kerb now". In London it stays on for six seconds. Then it goes out, or starts to flash, and a second clock starts. That second clock is the sum. The width of the road divided by 1.2. On a normal road that is six more seconds.
Those six seconds are for one person: the one who stepped off the kerb at the very last moment of the green man. The rulebook says this in its own words. The time after the green man "allows people to clear the crossing if they step off the kerb as the green symbol goes out". So the six seconds are meant to get that last person across. They do, if that person walks at 1.2.
Press the grey button and both dots step off the moment the green man comes on. Both get across, with time to spare. That is the other half of the story. If you get to the kerb early and go at once, the numbers are kind to you. If you get to the kerb as the green man is ending, which thousands of people do at every crossing every day, you get the countdown and nothing else. At 0.8 metres a second on a seven-metre road, the countdown runs out when you are still two metres from the far kerb.
In the 1950s, road researchers in Britain stood by crossings and timed people walking across. Then they picked a speed that 85 in 100 of those people could beat. It came out at four feet a second. In metres, 1.2. That was a fair way to pick a number. Most people, most of the time, walk faster than that. The trouble is who was crossing roads in the 1950s, and who is crossing them now.
America put four feet a second in its rulebook in 1961. Britain's rulebook, called the Traffic Signs Manual, still says 1.2 is the usual number. Germany says 1.2. Australia says 1.2, and adds that 85 in 100 pedestrians walk faster, which is true of pedestrians in general and not true of pedestrians over 65. Japan uses 1.0, and only counts to the middle of the road, because in Japan you are expected to stop on the island in the middle. India's code says 0.95, and says the cars must be kept on red for at least 12 seconds for every seven and a half metres of road. America cut its number to 1.07 in 2009 and kept it there when it rewrote its rulebook in 2023.
The engineers knew early on that the number left people behind. Their own handbook said so in 1965: about 35 in 100 pedestrians did not reach four feet a second. A study in New York found that at 1.2, half of all pedestrians would have to walk faster than they normally do. A study in Florida, at a crossing used by older people, found that 0.8 fitted 87 in 100 of them. The number stayed. Every second given to people on foot is a second taken from the cars, and the sum was written to keep the cars moving.
| where | the speed it assumes | what that means on a 7 m road |
|---|---|---|
| Britain | 1.2 m/s | 6 seconds after the green man. Councils have been allowed to use 1.0 since 2019. Few do. |
| Germany, Australia, Singapore | 1.2 m/s | 6 seconds. Australia allows 1.0 where more than 15 in 100 of the people crossing are children or older people. |
| United States | 1.07 m/s | 7 seconds of WALK, then 7 of flashing hand. It was 1.22 from 1961 to 2009. |
| Japan | 1.0 m/s | 7 seconds, but only counted to the middle of the road. You are expected to stop on the island. |
| India | 0.95 m/s | 8 seconds, and the cars kept on red for at least 12. |
Sources: Department for Transport, Traffic Signs Manual chapter 6 (2019), sections 11.7 to 11.9; the American rulebook, MUTCD, sections 4E.06 (2009) and 4I.06 (2023); LaPlante and Kaeser, 2007; Transportation Research Record 1438, 1994; the German rulebook RiLSA; Austroads via Main Roads Western Australia; NCHRP international practice; IRC:103-2012.
The speeds engineers quote come from healthy volunteers walking a marked line in a lab. Healthy people in their seventies do about 1.2 in a lab. If everyone were a healthy volunteer, the number would be fine.
In 2005 the Health Survey for England sent nurses into people's homes. They visited 3,145 people over 65, chosen at random, and timed them walking eight feet at their normal pace. Twice. The men averaged 0.9 metres a second. The women averaged 0.8. Three in four men and five in six women were slower than 1.2. Some could not do the walk at all. Count them too and it is 84 in 100 men and 93 in 100 women. The researchers ended their paper with one sentence: the walking speed assumed for pedestrian crossings, 1.2 metres a second, is not right.
A bigger study followed 10,249 people over 60 for twelve years and timed them again and again, 31,015 times in all. One timing in ten was fast enough for the crossing. At 60, 18 in 100 people could cross in time. At 80, 3 in 100. The richest fifth of them, 20 in 100. The poorest fifth, 3 in 100. So the crossing is timed for the young, and it is also timed for the well off. Nobody decided that. It is what comes out of the sum when you put real people into it.
The newest study, from 2025, looked at 1,110 older people who have some trouble walking, the kind of people who use a stick or hold the rail on the stairs. Seventeen of them could walk at 1.2. That is 1.5 in 100. Their average speed was 0.77 metres a second, slower than the slow dot on this page.
| who | speed | where it comes from |
|---|---|---|
| The rulebook | 1.20 m/s | Traffic Signs Manual, 2019 |
| Healthy men in their 40s, in a lab | 1.43 m/s | Bohannon and Williams Andrews, 41 studies, 23,111 people, 2011 |
| Healthy women in their 70s, in a lab | 1.13 m/s | same |
| Men over 65, at home, England | 0.90 m/s | Asher, Aresu, Falaschetti and Mindell, 2012 |
| Women over 65, at home, England | 0.80 m/s | same |
| Older people who have trouble walking | 0.77 m/s | Age and Ageing, 2025 |
Children are a different problem. Ten-year-olds walk as fast as grown-ups. What they cannot do yet is judge a gap in traffic. In a pretend road at the University of Iowa, children only got as good as adults at choosing when to step out between cars at the age of 14. The countdown cannot help with that. It is the reason crossings exist in the first place.
Queens Boulevard is the widest street in New York. Up to 225 feet across, twelve lanes, sixty thousand cars a day. In the early 1980s the city's health department put every pedestrian death on a map and found a cluster: 22 deaths and 18 near-deaths along two and a half miles of it. Then they looked up the ages. Every one of the 20 people with a known age was 60 or over.
The fix cost about $150,000. Longer crossing times. A traffic light on each island in the middle, so a person who could not make the whole width could stop halfway and wait for the next green, without breaking any rule. Bigger speed-limit signs. Talks at old people's centres. Over the next few years the traffic on the boulevard went up 19 in 100 and deaths went down 43 in 100. People had called it the Boulevard of Death. In 2017 nobody was killed on it.
The island in the middle is the oldest trick, and it still works best. It does not change the sum. It makes the road the sum is applied to half as wide. Japan builds its whole rule around it. India's code requires it on any wide road. The dot you just watched stopped on one.
Later, New York did something else across the whole city. At 2,869 crossroads the walk sign now comes on seven to eleven seconds before the cars get their green. Nobody gets more time to cross. They get a head start, so that they are already in the road, where drivers can see them, before the first car turns. Columbia University looked at 6,003 crossroads from 2013 to 2018, with and without the head start. Where there was a head start, a third fewer people on foot were hurt. In daylight, deaths fell by 65 in 100.
Toronto tried countdown clocks instead, at 1,965 crossroads. People on foot liked them. Drivers could read the same numbers, and sped up to beat them. Crashes between cars and people went up 26 in 100.
Sources: Centers for Disease Control, Morbidity and Mortality Weekly Report, the Queens Boulevard Pedestrian Safety Project; New York City Department of Transportation, Pedestrian Safety and Older New Yorkers; Nature Cities, July 2025; Richmond and others, Injury Prevention, 2014.
Why not just use a slower number? Because a crossing is a slice of a loop, and the loop belongs to the cars. A traffic light runs a loop, usually one to two minutes long. Cars one way, cars the other way, then people. Every second you add to the people's slice comes out of a car's slice, every loop, all day. A lane lets through about one car every two seconds of green. One extra second for walkers on a two-lane road is one car per loop. With a loop every ninety seconds that is forty cars an hour, at each side of each crossing. Multiply by six thousand crossroads and that is the whole argument against a slower number.
In 2009 the Mayor of London asked Transport for London to make the traffic flow more smoothly. Transport for London went through the timings at 1,003 sets of lights in a year. Before it did, it tested the idea at nine crossroads with a proper study. The green man was cut from eight to ten seconds down to the national minimum of six. The countdown after it stayed at 1.2. Cameras watched 26,431 people cross. 1,200 were asked questions.
Nine London crossroads, January and February 2009. TRL report PPR411, summarised by Transport for London's road safety unit, October 2009.
More cars got through, six and a half in 100 more. More people started crossing on the red man: 46 in 100 before, 51 after. Fewer disabled people said they felt safe: 70 in 100 before, 58 after. Sixty in 100 of everyone asked did not know what the blank period after the green man meant. One person who was walked round the crossings by the researchers said: "It was quite fast and I only got halfway across before the green man went off." The report to Transport for London's board the next year said the year's work had been done "with no net disbenefit to pedestrians".
There is a catch on the other side too. People will wait about 30 seconds for a green man. After that they cross on red. That number has been in the research since the 1960s and every study since has found it again. Give walkers a longer green and the next lot of walkers get a longer red, and more of them step out in front of cars. There is no free second in the loop. That is why the cities that fixed anything did not change the number. They went round it.
One more thing. The Department for Transport has rules for how much a minute of a person's time is worth. Those rules say a minute spent waiting on foot is worth twice a minute spent sitting in a car. On paper, the walker's time is the expensive kind. The computer that runs the lights is set to keep the cars' delay as small as possible. Until the middle of the 2010s, nobody even measured how long the walkers waited.
| speed used | quiet crossroads | busy crossroads | full crossroads |
|---|---|---|---|
| 1.07 m/s | no real change for cars | a little more waiting | a lot more |
| 0.91 m/s | no real change, or a little | more waiting | a lot more |
What a slower number costs the cars, worked out for six real American crossroads. AAA Foundation for Traffic Safety, 2007, table ES-1. Most crossroads are quiet most of the day.
Since 2009, anyone over 60 in Singapore can tap their bus card on a reader on the pole. That crossing then adds between three and 13 seconds to the green man, for that one green only, depending on how wide the road is. Most add about six. Nobody else waits longer the rest of the day. More than a thousand crossings have the reader now, and the plan is half of the crossings in every housing estate by 2027. Hong Kong copied it in 2018, with the card people use for the metro. It is at 21 crossings, chosen near clinics and old people's centres and away from the busiest roads. Singapore's own study found that even with the card, people over 70 could not make one crossing in seven when they were carrying shopping. The study's answer was the island in the middle.
Japan counts its 1.0 metres a second only to the middle of the road and puts an island there. India requires an island on any wide road. New York put lights on the Queens Boulevard islands. It is the same idea everywhere. Do not make the sum cover the whole road. The slow walker gets two short crossings instead of one long one, and nobody has to pick a new number.
A puffin crossing, which Britain has had since 1992, has a sensor pointed at the road. If someone is still on the crossing when the countdown ends, it holds the cars a bit longer, up to a limit. The green man on a puffin is short on purpose, because the machine is supposed to watch you instead of guessing. The catch is the limit, and the fact that most crossroads in a city do not have the sensor. In November 2025 the government's guidance said a crossing timed at 1.2 with no sensor only counts as a safety problem if you would have to walk faster than 1.2 to get across. That is most people over 65.
Eighteen crossings in London have been turned round since 2021. The green man is on all the time. It only goes red when a car is coming. People cross without asking. More people obeyed the lights, 13 in 100 more, because there was nothing to disobey. Transport for London called the trial a success and said it would do more where the roads suit it. It is a small number of crossings. It is also the only fix on this page that does not start from the cars' timetable.
And there is the Barnes Dance, which is what New York called it in 1962 when all the cars were stopped at once and everyone crossed in every direction, corner to corner. Kansas City and Vancouver did it first, in the late 1940s. Shibuya in Tokyo has the most famous one, three thousand people every two minutes. Oxford Circus got one in 2009. The dot on this page walked its diagonal.
You would think the timings were public. They are set by public bodies, on public roads, for public safety. In most places they are not written down anywhere you can see. In Sydney a software engineer called Jake Coppinger got fed up waiting at a crossing near his flat, could not find out how long the wait was supposed to be, and built a website where anyone can stand at a crossing with a stopwatch and type in what they timed. 148 crossings so far. The longest wait anyone has recorded is 179 seconds. The state still does not publish its timings, so his numbers are the only public ones there are.
The road you just watched, Botany Road, was timed ten times by ten different people. The green man averaged 7.8 seconds. The flashing red averaged 13.6. The rule would have given 13. So the real light is a little kinder than the rule, and on a narrow road, if you step off when the green man comes on, a little is enough. The slow dot made it with about two seconds to spare. It is the only road on this page she crossed without stopping.
Then the red man stays on for ninety seconds. People wait about thirty.
The same seven-metre road, the same six seconds plus six, and everyone who comes to it. Each dot walks at a speed taken from the studies above: lab speeds for people under 65, the England survey for people over 65, mixed the way the country is mixed. They come to both kerbs, wait, and step off between half a second and two and a half seconds after the green man, the way people do. When the cars get their green, anyone still in the road turns orange, and the car in that lane waits for them.
Ten real roads, at the average speed for healthy people your age measured in labs (23,111 people, 41 studies), and, if you are over 65, at the speed the England survey found people your age walk at home. You step off one second after the green man comes on. The lab speed is the kind case.
It depends where you are standing. There are three kinds of crossing, and the same button is on all of them.
A quiet road. Nothing happens until someone asks. Press it and the cars get red within a few seconds. On a puffin a sensor also checks you are still there.
About 3,000 of London's 6,000 signals. The green man has a fixed slot in the cycle. Pressing puts in a "demand" so the slot is not skipped when nobody is waiting. It cannot come sooner or last longer.
The green man runs every cycle whether you press or not. New York counted its own in 2004: 2,500 of the 3,250 buttons did nothing. Regent Street's works between midnight and 7am.
The one button that always does something is the card reader in Singapore and Hong Kong, because it does not ask for a green. It asks for a few more seconds.
New York Times, 27 February 2004; BBC News Magazine, 4 September 2013; Transport for London.
Britain's rulebook has allowed 1.0 since 2019, at any crossing a council chooses, or everywhere in an area. Most councils have not used it. In July 2023 the government's walking and cycling body said it was thinking about making 1.0 the recommended number for England. On a normal road that would mean seven seconds instead of six. Living Streets, the walking charity, has been asking for three more seconds since 2013, and welcomed it. In November 2025 the same body's guidance still said 1.2 is what most crossings in England are timed to.
America's number is 1.07 and has been since 2009. An American study in 2006 took the walking speeds of each age group and the forecast of how old the country will be, and worked out that by 2045 the slowest 15 in 100 walkers, the people the number was always meant to protect, will do 1.09 metres a second. That is below the old number. People got older. The number stayed where it was.
In New York, people over 65 are fewer than 15 in 100 of the population and more than 45 in 100 of the pedestrians killed each year, and that share has not moved since the city started its road safety programme. In Britain, 61 in 100 of the pedestrians killed are over 60. Most of those deaths, about three in four, happen more than 50 metres from any crossing. The crossing is the safest part of the road. The sum decides who feels able to use it.
The sum. Clearance = crossing length ÷ design walking speed, rounded up, from the DfT Traffic Signs Manual chapter 6 (2019), sections 11.7 to 11.9, with London's six-second green man from TfL's answer to FOI request 2974-2324. The rulebook road is the manual's own worked example: two lanes of 3.5 m.
The two dots. 1.2 m/s is the design speed. 0.8 m/s is the mean walking speed of women over 65 in the Health Survey for England 2005, timed over eight feet at normal pace at home, 3,145 adults over 65, published by Asher, Aresu, Falaschetti and Mindell in Age and Ageing, 2012. In the button scene both dots step off as the green man ends, which is the case the rulebook says the clearance is designed for, or at once if you press the grey button. In the chapters, the caption says which.
The roads. Widths are OpenStreetMap crossing ways measured node to node on 7 September 2026: Oxford Circus diagonal 27.7 m (way 622158403), Aurobindo Marg at AIIMS, Delhi, 41.4 m (682848213), Lorong 2 Toa Payoh, Singapore, 26 m (1553550198), Botany Road, Sydney, 14.7 m (1198615528). Queens Boulevard at 63rd Drive is 200 feet, 61 m, with two medians, from the published width. Timings are each country's rule applied to that width: Britain 6 s plus L ÷ 1.2; the United States 7 s WALK plus L ÷ 1.07 (MUTCD 4I.06); India L ÷ 0.95 with traffic held red (IRC:103-2012); Singapore 24 s of green man, the middle of the 21 to 28 s observed on crossings of 22 to 28 m in a 2011 field study by the Yong Loo Lin School of Medicine, plus about six seconds for Green Man+ (Land Transport Authority); Sydney 7.8 s green and 13.6 s flashing red, the mean of ten stopwatch measurements at Botany Road on Better Intersections, open data under ODbL. A rule-derived timing is the rule, not a stopwatch. A real signal can be set longer.
Everyone. Speeds for people under 65 are the decade means from Bohannon and Williams Andrews, Normal walking speed: a descriptive meta-analysis, Physiotherapy 2011, 41 studies, 23,111 adults, with a spread of 0.2 m/s from Bohannon 1997. Speeds for people over 65 are the England survey means, 0.9 for men and 0.8 for women, with the spread that puts 76 and 85 per cent below 1.2. The age mix is the UK adult population by decade from ONS mid-2023 estimates. People step off 0.5 to 2.5 seconds after the green man. Half of the people who arrive during the countdown go anyway, which is roughly what TfL's countdown trial observed. Cars are 4.4 m long and pull away at up to 7 m/s. It is a model, and it is on the generous side: real people carry shopping.
The rest. The 1965 handbook estimate and the New York and Florida findings: Transportation Research Record 1438, 1994. The 1961 origin of 4 ft/s: LaPlante and Kaeser, 2007. The 1950s origin of 1.2: the Department for Transport, via BBC News, 24 July 2023. The twelve-year study: Webb, Bell, Lacey and Abell, Preventive Medicine, 2017. The mobility-limited study: Age and Ageing 2025, afaf345. Children: Plumert and Kearney, University of Iowa, 2017. London 2009: TRL report PPR411 via TfL's London Road Safety Unit research summary 16, and TfL's Surface Transport Panel paper of 19 May 2010. The cost to cars: AAA Foundation for Traffic Safety, 2007. The value of waiting time: DfT TAG unit A1.3. The 30-second wait: TfL literature reviews. Queens Boulevard: CDC MMWR, 1989; New York Times, 2017. New York seniors and head starts: NYC DOT; Nature Cities, July 2025. Toronto: Richmond et al, Injury Prevention, 2014. Singapore: LTA, Ministry of Transport, and the 2011 field study. Hong Kong: Transport Department. London green person authority: TfL, 2021 to 2023. Buttons: New York Times, 27 February 2004; BBC, 4 September 2013. Britain's deaths: RoSPA, 2023; DfT Reported Road Casualties 2025. Where it stands: DfT Traffic Signs Manual §11.7; BBC, 24 July 2023; Active Travel England, 6 November 2025. The 2045 forecast: Fitzpatrick, Brewer and Turner, 2006.
The numbers used on this page: crossing.json.
Drop your email. The next one lands free. Nothing on this page is locked.