Running across Westminster Bridge

How long does it take to run across Westminster Bridge?

This question was inspired by an article in the Times on 6 November 2018. It reported that a twitter post had gone viral. A man named Scott Pack had commented on a long-forgotten feat performed by his grandmother on 14 April 1934. Apparently it used to be a game among people working in the UK’s House of Commons to try to run all the way across Westminster Bridge during the time Big Ben took to strike noon. According to Pack, his grandmother Florence Ilott had been the first person ever to achieve it.

It may have been a prominent feat back in the day but Ilott’s feat and name had been lost to history till Pack wrote his post. Some fairly obscure figures who were once famous have been lauded in the Times obituaries, such as Billie Dovey or June Robles. But Ilott? I don’t remember reading about her.

The question was, was the feat plausible? Westminster Bridge doesn’t seem overly long and Big Ben tolls remarkably slowly but that didn’t mean the tale was true.

The claims for Ilott’s feat aren’t entirely consistent. Newsreel footage claims she ran 353.6m. A written account says that she ran from the steps of St Thomas’ Hospital to Big Ben. That’s an east to west run and the Times says it’s about 350m. Another account says she ran 230m. I am deliberately giving all the measurements here in metric.

It’s also worth saying that the press accounts say she made it by the stroke of ten. Ilott, who died in 2002, only told her grandson that she’d done it before the stroke of twelve.

Exactly how long is Westminster Bridge? One length given online is 252m. Wikipedia says 250m but that sounds like a rounded figure so I shall assume 252m is correct.

On the basis of these details we can disregard the 230m figure and we have three possible distances: 252m, 350m and 353.6m.

Next we need the timings. How long does it take Big Ben to strike ten, and how long does it take to strike twelve?

There are plenty of recordings of Big Ben on YouTube but they don’t offer a consistent strike rate. There is one video labelled “Big Ben strikes 12 (Good quality sound)” in which the time taken from the start of the first stroke to the start of the tenth is fully 39.3 seconds. Other recordings take only a fraction over 37 seconds to play the same strokes.

Which means that either Big Ben’s strike rate is variable or else some of these YouTube videos are not as reliable as they should be.

We’re going to assume a consistent strike rate and now the UK parliamentary website comes to our rescue. At there is a handy set of downloads of the strikes. Analysis of the MP3 of Big Ben striking twelve suggests that each stroke can be heard for roughly three seconds, and that:

Time to start of tenth stroke: 37.2 seconds

Time to start of twelfth stroke: 45.5 seconds.

So, is Ilott’s feat plausible? Let’s look at the times in metres per second:

252 metres 350 metres 353.6 metres
37.2 seconds 6.77 m/s 9.41 m/s 9.51 m/s
45.5 seconds 5.54 m/s 7.69 m/s 7.77 m/s

How does that compare with top athletes?

There’s no 250m or 350m athletics event so our closest points of comparison are the 200m and 400m. The women’s 200m world record is 21.34 seconds but that was set by Florence Griffith Joyner and is considered problematic. The women’s 400m world record is 47.60 seconds.

World records are not a realistic benchmark in any case unless we are trying to prove that something is impossible. Olympic qualifying times should give a more realistic view.

For 2016, the 200m qualifying time for women was 23.20 seconds, giving a minimum average speed of 8.62 m/s

For 2016, the 400m qualifying time for women was 52.20 seconds, giving a minimum average speed of 7.66 m/s.

All of which suggests that crossing Westminster Bridge by the twelfth stroke or even by the tenth is plausible. Ilott may have made the longer distances by the stroke of twelve but that’s a much tougher feat.

The distance of 353.6 metres is derived from newsreel footage that gives the distance as 1160 feet. The footage is below. You only see a little bit of her running but it’s clear she could run very fast indeed.

There is an interesting follow-up to the Times article. A letter to the editor in the 9 November edition of the same paper said that students at St Thomas’ Hospital in the 1960s would often try to run all the way across the bridge while Big Ben was striking midnight after having spent the night drinking scotch ale in the students’ club. And, said the correspondent, most of them made it… despite being dressed in leather shoes, blazer and flannels.

The Melancholy Roman

What the Economist doesn’t realise

The article heading from this week’s Economist

There’s an article in this week’s Economist titled “Stand clear of the doors.” The subheadings make it clear that the article is about making trains run on time and more specifically, about using artificial intelligence to assist with managing the platform-train interface. The image above is part of the screenshot of the article on my iPad.

The article talks about congestion and issues surrounding people getting on and off trains, or getting too close to the platform edge. It suggests that artificial intelligence can be applied to this problem. With the many cameras on station platforms and inside carriages, it is possible to analyse the images from these cameras to predict where congestion will occur.

Sounds reasonable, right? And then, as the article says, passengers can be directed away from the congestion, for example to less crowded doors. And there’s a team at the University of Lancaster working on this particular problem.

But here’s the detail the Economist seems unaware of: station staff already know where congestion will occur.

I should give some disclosure at this point: I worked for London Underground for ten years.

Human beings are creatures of reasonably consistent behaviour. They tend to do similar things for similar reasons. And thus, for example, on platform 8 at Baker Street Station on a busy Saturday afternoon, you can put money on the fact that the first two train carriages will be very crowded indeed. The last two carriages of the same train will have empty seats.

An eastbound Piccadilly line train departing South Kensington Station will have a lot of people in the middle part of the train, especially if it has been held in the platform for any length of time. The middle is where the platform entrance is.

I could easily give dozens of examples like this, and this brings me to the second detail that the Economist seems unaware of: it’s really difficult to get people to move along a platform.

Regular commuters know where the best place to wait is. I used to live between Perivale and Alperton Stations in west London. If I caught the Central line home from central London I would get on the front of the train. If I caught the Piccadilly line I would get on the back. In each case I was positioned for the door closest to the exit when I got off.

Persuading people to move along the platform when they know where they want to stand is a nigh on impossible task. Heck, persuading them to move along even when they don’t know where they want to stand is hard. I just can’t see the technology working in the way the Economist describes.

So why has the rail industry expressed an interest in the idea? Well, the article mentions that the team from the University of Lancaster has set up a company called Digital Rail. Their website gives a clue.

The artificial intelligence technology that they’re producing won’t just be useful for predicting congestion. It should also be able to generate alerts when certain events happen. Among those events are people falling in the gap at the platform edge, and people falling or jumping in front of trains.

People falling in the gap is a serious issue. It’s why the northbound Northern line platform at Embankment Station is permanently staffed and why a bunch of other stations with curved platforms have staff on them at times of day when statistics show that accidents are likely to happen.

But people falling or jumping in front of trains is the biggie. Some lines could already run driverless and others will have the capability once the signalling systems have been upgraded. But for safety’s sake there has to be a way to stop a train when someone has gone down in front of one and the technology just isn’t there yet.

Artificial intelligence will probably solve this problem and when it does there’ll be no need for train drivers anymore. That’s the end goal.


The Melancholy Roman