The Ninety Second Railway: Making the Victoria The Most Frequent Metro in the World


Back in the 1990s, John Self, the general manager of the Victoria line at the time, gave a talk about his vision for its future. From the outset he pulled no punches. It was time, he said, to start thinking about what was going to happen when the line’s trains needed replacing. It was time to start planning for a Tube line fit for the 21st Century. He called his talk ‘The Ninety Second Railway’.

The title was a reference to what he believed the line would eventually need – trains that ran ninety seconds apart. Or, to put it another way, he was proposing to run 40 trains per hour (tph) on the Victoria. This was because he was convinced that Britain’s railways were not merely starting to show signs of arresting and reversing their longterm decline in usage. He saw the increasing passenger numbers as a trend that would continue for the foreseeable future. As a result, he predicted that demand would increase significantly on the Underground in the coming years.

The talk was particularly well attended. In addition, many of the attendees were people with detailed knowledge of the complexities of running railways and were familiar with the Victoria line and its particular issues behind the scenes. The audience listened attentively, but it was not at all clear how well the proposal was going down.

The talk ended and it was time for questions. Thanks to the makeup of the audience, these were of unusually high quality and particularly thought-provoking.

Whatever the question, it was accompanied by a general enthusiasm for the ideas put forward. This always came, however, with a caveat – that although such thinking should be encouraged, perhaps the speaker should set his sights a little lower.

There wasn’t complete agreement as to what ‘lower’ meant in this context, but the general feeling was that getting to 32 or 33tph would be an achievement in itself. Some were bold and suggested aiming for 36tph, but at the same time, they cautioned that this was something to aim for, rather than plan around on the assumption it was achievable.

Ultimately, the cognoscenti within the audience seemed to fall into three distinct groups.

Those who knew what had happened

There were those that had seen this all before. These tended to be older men (and they were all men), often retired, who had been heavily involved in the design and building of the Victoria line during the 1950s and 1960s. They were aware of how difficult it had been to achieve what was already in place.

They probably also had a very good idea of what was preventing more from being achieved. To them, the Ninety Second Railway seemed impossible with the infrastructure that existed already.

Those who knew what was possible

There were those that understood what could now be achieved. Generally, these were the modern railway technologists – people whose day job meant they knew the current limitations of technology. This was a time when the latest thing around was the 1992 stock on the Central line. It was also a period when automatic train operation (ATO) on both the DLR and the Central line took years to fully commission and one where cathode ray tubes (CRTs) were the only realistic way of displaying an image of the platform in the driver’s cab. At this point, the Victoria line had been running for over 25 years, entirely underground, yet its trains still couldn’t consistently stop at the same location on the platform without occasional driver intervention. To them, the Ninety Second Railway seemed impossible with the technology of the day.

Her Majesty, Queen Elizabeth II at the opening of the DLR

Those who knew what was affordable

Present too were the planners and financiers. They understood how practical it was to get financing for railway projects. The 1990s were not the time to be asking the government for large sums of money for rail schemes in the public sector. Who was going to pay for the extra trains required? For the alterations and expansion necessary in the depot? For the new signalling system that would be required? There were probably those present who argued that getting the money for a like-for-like replacement of the existing rolling stock would be challenging enough.

A success story turned bad

For the first couple of decades, the Victoria line was generally viewed as a success – at least by those on the outside. As a construction project, it had set out with a clear goal and achieved it.

What no-one was really prepared to say, however, was that in one respect the Victoria line project had failed. It had not lived up to expectations with regards to frequency.

Many of the early technical reports (and a range of other sources) make it clear that the line was designed to be a 30tph railway – the ‘One Hundred and Twenty Second Railway’ if you like. Yet all the evidence seems to suggest that, apart from short unsustainable bursts at this frequency, it was quite unable to achieve this. Indeed it would not reliably and regularly achieve this until around 2013 – after the trains and the signalling system had both been replaced.

Why the line’s designers and builders ultimately failed to achieve this provides an insight into what it takes to run a modern Tube at such frequencies – or even at higher ones.

A different world to today

Detailed design of the Victoria line took place in the 1950s. It was a different world. The transistor had only recently been invented (1948) and integrated circuits did not yet exist. ‘Computers’ were extremely common, but not in the way one thinks of them today. At the time, the term generally referred to a person who computes. The electronic variety was very rare and very expensive.

The people who designed the Victoria line were forward-thinking – indeed almost recklessly so. Even with the advance of technology during the construction phase, the electronics were incredibly crude by modern standards when the line opened in March 1969. To put this in context: four months later, man first landed on the moon with only the most basic of computers to assist them.

The limitations of technology meant that the means of controlling a Victoria line train’s speed were crude in the extreme, by modern standards at least. Acceleration was unrestrained, planned coasting and braking was determined by emitted signals at fixed positions on the track and there were a limited range of additional functions to further restrict performance if there was a train ahead. Ultimately this control system probably delivered a better performance than a bad driver, but a worse performance than a good one.

The original rolling stock, pictured at Tottenham Hale. Courtesy Oxyman

It was not just the signalling system that relied on the latest technology. The line was the first to have Driver-to-Controller communication (at least most of the time) by means of a carrier wave system. The leaky feeder method of propagating radio waves in tunnels had yet to be invented and was still some years off.

More fundamentally, in the middle of the 20th Century, an electric motor generally meant a direct current (DC) motor. The advancements which led to more sophisticated, more controllable and powerful AC motors in electric trains had yet to take place.

An imperative to reduce staff costs

Despite being hampered by the technology of the day, the planners were also forward-thinking in other related areas. Recognising the need to obtain government approval for the cost of the line, it was seen as imperative that the trains could be operated by just one man (and, again, in those days, it would have been a man). This was intended to reduce operating costs and increase reliability. Moreover, as the 1960s began, it was becoming increasingly difficult to recruit staff for the Underground. Simply paying staff more money was not a realistic option – not least because pay restraint was a key concern of the government. “One man’s pay rise was another man’s price increase,” as Prime Minister Harold Wilson famously said.

Real world insight

It was never never envisaged that all trains would go to the northerly terminus at Walthamstow (Hoe St). Indeed Walthamstow Central (as it would become) was not expected to handle more than around 20tph – well within the capability of a two track terminus. As not all trains would go to Walthamstow Central, and because there was a need to provide access to the depot, a turnround facility was provided at Seven Sisters for trains terminating short. This was relatively well-designed, with a separate platform for terminating trains, but – as we shall see – it wasn’t quite designed well enough.

Meanwhile down south, it must have been immediately apparent to planners that a critical restriction on the line’s frequency would be its ability to achieve the necessary throughput at its southern terminus. Initially, this was Victoria, but the line was always intended to continue further south. The destination finally chosen was Brixton.

It helps a lot when designing something if you can look at a precedent. When it came to Brixton station and the track layout, there weren’t a lot of twin track, dead-end terminals underground to learn from. At the time, apart from the Waterloo & City line at Bank, then run by British Rail, there was only the Northern City line terminus at Moorgate – and the southern Bakerloo line terminus at Elephant & Castle.

Elephant & Castle: the best role model available

The Bakerloo’s southern terminus at Elephant & Castle must have attracted a great deal of attention from the Victoria’s designers. It was certainly the best terminus available for them to draw lessons from. It was also intensively used, as it had to handle all trains on both the Stanmore and the Watford Junction branches of the Bakerloo line. It was just about able to handle 30tph, but even then it was struggling to cope all those train movements. It may be hard to believe now, but 50 years ago the Bakerloo line was the busiest Tube in central London. If the Bakerloo could have handled more trains, there is little doubt that it would have done so. Relief to the Bakerloo line would eventually come in the form of the Jubilee line, which took over the Stanmore branch – a branch which the Bakerloo had itself inherited from the Metropolitan line.

Elephant & Castle was clearly a post-war limitation on the service the Bakerloo line could provide. Indeed at the time, the London Passenger Transport Board (LPTB) were keen to extend the line from Elephant & Castle to a proposed three track terminal at Camberwell Green in order to overcome this. The new terminus would also have brought the added benefit of a long-sought-after station at Camberwell.

What the terminus at Elephant & Castle desperately lacked, in order to provide a useful comparison for the Victoria line, was passengers. This isn’t to say that there weren’t any. Simply that they were not present in anything near the quantity expected on the Victoria line then (let alone what it sees now). This is because, even then, the Bakerloo had largely acted as a connection between the suburbs of north London and central London itself. Southbound, then as now, the majority of passengers would have alighted by the time trains passed Waterloo.

Indeed the Bakerloo was only really extended from there to Lambeth North station in order to reach the depot at London Road, close to St George’s Circus. Having reached that far, it made sense to take the line a little bit further to Elephant & Castle and at least try to tap into the limited extra traffic this would bring. This lack of passengers explains why, practically since opening, there have always been proposals to extend the Bakerloo line southward.

A schoolboy error?

With a lack of passengers in their best example, it is hard to see how the Victoria’s planners could accurately factor the human element in their calculations.

This isn’t to say that they didn’t try. Various exercises were carried out involving simulated crowds. Unfortunately, these made the schoolboy error of… well… using actual schoolboys – a group of people who can hardly be said to represent a meaningful cross-section of Tube travellers. The line’s planners were not blind to this, and there is sufficient evidence to show that they attempted to adjust their data to reflect typical Tube users. A more representative group of test subjects, however, would likely have been more useful. Perhaps it was just too difficult to organise or, perhaps more likely, it was deemed too expensive to conduct the trial with more appropriate stooges.

Another possible reason for overconfidence, later on, was that a lot of reliance seems to have been placed on experience gained elsewhere – with the 4-car Victoria line trains that ran an early shuttle service between Woodford and Hainault on the Central line.

This may well have provided the planners with a lot of real world experience on the engineering side of things. However, it was also a lightly used open-air section of the Underground serving five stations and including what was then the least-used section of the network. As a result, it was hardly going to provide the stress testing needed to ensure you can run 30tph.

Getting things right

This is not to say that the Victoria line planners didn’t get a lot of things right. The termini at Walthamstow Central, Victoria (temporarily) and Brixton had long crossover tunnels outside of the stations. These enabled a fast approach to the platforms. The stations also had long overrun tunnels, so there was no need for the signalling system to restrict the speed of approach (in case the train failed to stop correctly). Remember that they were relying on the train stopping automatically and sufficiently accurately – and they had already factored in having all platforms being longer than the trains to cater for any resulting margin of error.

In addition to the attention to detail paid to the track, the performance of the trains themselves was not neglected. Unfortunately, in those days, all that could be done to improve performance was to have more motored bogies. This was done, but one can only go so far with it. This is because, with each powered bogie added, the benefit, as measured in time-saving, goes down, whilst the weight penalty of each additional bogie remains constant.

The Brixton loop

It isn’t clear whether the planning for a fast terminal turn-round time was deficient, whether the measures in place simply turned out not to be enough or whether they were sufficient for 30tph but other factors restricted frequency. Whatever the case, the first proposal to build a loop at Brixton via Herne Hill (where there would be a single platform station) was not long in coming. This in itself seems to suggest that, at one point in time at least, the terminus at Brixton was the primary limiting factor to frequencies on the line.

The suggestion, taken quite seriously in the 1980s, was to have a loop at Brixton, rather like the Kennington loop on the Northern line, so that trains could continue from a southbound platform and reappear in the northbound platform without having to reverse direction. As Piers Connor reports, the idea was abandoned for two reasons. The first was that it would have required two extra trains. The second was that the crowding caused by the expected number of passengers added by a loop station at Herne Hill would have eclipsed the relief benefit provided by the loop itself. This was hardly the desired result.

Interestingly, it seems that no consideration was given to providing a loop at Brixton without a station at Herne Hill. This would have probably have obviated the need for a least one of the extra trains, but the relief gain would have been small in proportion to the still-large tunnelling and fit-out costs. Without the up-tick of passenger fares a station at Herne Hill would have provided, this meant it would probably have failed any business case or financial scrutiny.

The theory

Aside from the loop, Piers Connor also reports on frequencies achieved on the Victoria line in the 1980s. In particular, in part 6 of his articles on the Victoria line in Underground News (sadly not all parts are available online) he states:

From 1975, the Victoria Line service was increased to normally run with 34 trains during peak hours, although this dropped back to 33 between 1979 and 1985. The service was generally based on a 126-second headway, with a 120-second headway during the 20-minute ‘peak within the peak’.

Now, as Sir Humphrey would say, these figures are liable to be misinterpreted.

Firstly, there never was a 126-second headway. The Victoria line, for most of its life, has run to a timetable specified to the nearest quarter of a minute. 126 seconds equates to approximately 28.5tph. In reality, it was a timetable where roughly two out of every three trains at 120-second intervals whilst the third train had a 135-second interval. In addition, approximately two out of three trains would go to Walthamstow Central and one would terminate at Seven Sisters. Quite surprisingly, the timetables of the day suggest this is just a coincidence and the turnout at Seven Sisters did not cause a delay.

Secondly, it is worth pointing out that the ‘peak within the peak’ lasted for only 20 minutes. This was a short term burst, possibly only in one direction. It was clearly not sustainable or it would have lasted longer. A 20 minute peak on a line with an end-to-end running time of over half an hour naturally suggests some tweaking to the basic timetable to get a very slight improvement of frequency. Indeed in terms of capacity, it is not even equivalent to half an extra train during that 20 minute period.

What Conner does not make explicitly clear is that this frequency was not subsequently achieved in later timetables. In all probability, the increase in passenger numbers also increased dwell time and prevented the more frequent service. So it may well have been that later the limiting factor became the dwell time at Victoria station (northbound), especially as there was an initial reluctance to temporarily close the platform to people entering the station to reduce overcrowding.

The practice

Mike Horne, who was a manager on the Victoria line during his career, paints a very different picture from that given by Conner. It is certainly a less positive one. As he writes online:

The Victoria Line that I knew and for quite a while worked on was a pig of a thing. It was being pushed well beyond the expectations set for it when it was built and was running a 2 – 2½ minute service, or about 27 trains an hour, in theory. This was more than it was designed for and only the infiltration of 1972-stock cars off the Northern Line allowed the fleet to be topped up to provide sufficient trains (24 trains an hour was the most originally contemplated). The signalling and automatic train control systems did not work well with such intensity. The slightest irregularity would cause a train following too closely to just catch a 270 (restricted speed) code instead of a full speed code, and condemn it to run at controlled speed into following platform, perhaps losing it 10-20 seconds, depending on location. This cost capacity and reduced effective throughput. The track layout was hugely inflexible, especially at Seven Sisters, and any significant delay would cause mayhem with the service as the controllers and crew managers battled to sort things out. The railway was working at its limits, so when something went wrong it had a disproportionately big impact. The service either ran well, or was dreadful. It was hard to be in between.

In a modern day equivalent to John Self’s talk from the 1990s, the article is entitled 33 Trains an hour – but why not 40?

As Horne indicates, by the time the 1967 Tube Stock was withdrawn the Victoria line service was running at only 27tph. This was achieved by running the trains 135 seconds apart, regardless of destination. By now the line had 37 trains in service thanks to butchering some Northern line 1972 Stock, yet as can be seen, this was providing a worse service than had been delivered with just 33 trains 30 years previously.

The trains

When the time came to replace the Victoria line trains consideration had to be made as to how many units to order. The 1967 Stock consisted of 43 trains, of which 37 were generally in service during the peak. With better analytical tools than before available, it was concluded that a 33tph service was achievable. This was 10% better than the best achieved previously and around 22% better than was running at the time.

A further complication was that by now the Underground had entered the era of Public/Private Partnerships. The nature of these deals discouraged risk-taking – such as planning on running driverless trains or aiming for too high a frequency. This was because the financial risk of failure for the private contractor would rarely, if ever, eclipse the likely uplift in the farebox. Indeed this “play safe” attitude would ultimately lead to the private company responsible for upgrading the Victoria line (Metronet) ordering more trains than it really needed. After all, it was a cost that could be passed on to the ‘customer’ (London Underground) with few – if any – repercussions.

Nonetheless, this upgrade should at least have yielded 33tph on the Victoria line, then seen as the likely maximum, and that should have been the end of the story.


The first stage of that upgrade was simply (and we use that word advisedly) to replace the trains with the new purpose-built modern 2009 Stock and the signalling with a state-of-the-art Westinghouse (now Siemens) system called Distance-To-Go. During this period, the new trains were effectively ‘slugged’ due to external restrictions on power supply and the need to co-exist with 1967 tube stock which was still dependent on the old first-generation signalling.

The 2009 Stock at Seven Sisters. Courtesy Martin Addison

Once this work had been completed, it soon became clear that the planners of this generation had perhaps been as pessimistic as those of the previous generation had been optimistic. The signalling system was a world away from that which had originally been installed. Despite the DC power supply, the new trains were fitted with modern alternating current motors. One advantage of these is that they are ‘notchless’ – the rolling stock equivalent of being gear-free. It was the equivalent of doing 0-60 in an electric sports car as opposed to a petrol-engined one. In such a situation, the electric car will alway win, because one doesn’t need to go through gear changes.

A taste of things to come was apparent with the first timetable improvement. This was to 28.5tph, achieved without adding any trains to the timetable. 47 trains had been purchased, but the peak timetable needed only 37.


Things got even better. 30tph would provide a nice, steady pattern of trains 120 seconds apart through central London and this was the next goal. Any variation in this interval at the terminals would simply be compensated for, if necessary, by an appropriate short extra pause at the next station. The assumption (rightly) was that few passengers would notice the extra 15 seconds of dwell time that would result. Impressively, 30tph was also managed without requiring any extra trains in regular service. Understanding the context of what had come before, it should not be understated just how much of an achievement this was. Almost effortlessly, the new trains and signalling had already exceeded the best regular performances ever achieved on the line before.

The big one – 33tph

The really big change, however, was to push the frequency up to 33tph. It is important to note just how this was achieved. At the time, the working timetable was only specified to the nearest 15 seconds. An interval of 105 seconds would have given an even service pattern of over 34tph. This wasn’t, however, what was implemented. Instead, the timetable was altered to run four trains 105 seconds apart, followed by another 120 seconds later. The result was five trains every nine minutes – or the slightly awkward 33.3tph.

That London Underground were hedging their bets on the new timetable is clear from the publicity that announced its launch – or rather, the lack of it. Only after it had been in operation for two weeks did they issue a press release announcing this improved level of service.

The push for more

The successful introduction of 33tph on the Victoria line brought with it an obvious question – could the timetable support even more trains?

It would seem obvious that if you can run four trains 105 seconds apart in succession, then you ought to be able to run five. Why this was not thought possible is a bit of a mystery. In the end, the extra time allowed for the fifth train was removed in 2014. This gave the line a peak hour service of 34.3tph. Despite the slight increase in frequency, the service still only used 39 trains out of a possible 47 in the fleet. Again, bets were hedged. The press release confirming the new frequency was only issued a couple of weeks after it had been implemented.

The push to 36tph

By now it was clear that 34tph (rounding down) was about as much as the Victoria line could take without further major upgrade work. Not least because it would require a move away from a service pattern that regularly saw trains terminating at Seven Sisters.

That shift in service pattern was soon realised to be key. It was a key, however, that could unlock an unprecedented service not just for the Underground but, potentially, the world.

The work at Walthamstow

How those upgrades were completed could form the basis for a major story in itself. The short version, however, is that a working party determined that it was both possible and highly desirable to end the practice of terminating roughly every fourth train at Seven Sisters. They recognised that whilst the old track layout at Walthamstow Central could not cope with this, the modern signalling now could. Also, that the scissors crossover at Walthamstow Central (one of the major limiting factors) was arguably overdue for renewal anyway. By upgrading the crossover and examining the detail of all aspects of Victoria line working, it was determined that 36tph throughout the length of the line was achievable.

Resulting track work

So it was then that on Monday 22nd May 2017 the Victoria line ran a peak hour service of 36tph. This continued for about 80 minutes, a feat that was repeated in the evening peak as well. A long-desired dream of London Underground and the men and women of the Victoria line had finally been achieved.

And beyond?

That is not the end of the frequency story on the Victoria. The plan is to increase the period of 36tph peak running from 80 minutes to around three hours in both the morning and evening peaks. After that, London Underground’s attention will shift to achieving 36tph on the central section of the Jubilee line. The aim is to have this in place by the end of 2020.

Table from paper on World Class Capacity Programme presented to Programme & Investment Committee

The obvious question, however, remains: can an even higher frequency be achieved?

36tph means trains are 100 seconds apart. The obvious next step would be to push for 95 seconds. This would translate to a frequency of around 37.9tph – or 38tph when rounding up.

Critically, there would appear to be sufficient trains to run the service – one of the few positive legacies, albeit an accidental one, of the PPP. The current timetable requires 41 trains in service in the morning peak and, curiously, only 40 trains in service in the evening peak – so round trip times of 4100 and 4000 seconds respectively. If a round trip could be done in 4085 seconds then a 95 second interval services would require 43 trains in service. This would still leave 4 spares for maintenance.

Of course, having the trains is only one part of the requirement. As our editor John Bull is prone to point out, there comes a point where frequency is not about how many trains you can squeeze through the tunnels, but about how quickly you can get passengers onto and clear of, the platforms. Nonetheless, it is significant to note that senior people are daring to ask the question. Most notably, Network Rail Chairman (and ex-Transport Commissioner) Sir Peter Hendy, who spoke about it in his keynote speech on The Future of Britain’s Railways last year.

For now, then, last century’s dream of the Ninety Second Railway remains tantalisingly just out of reach. That we now have the hundred second railway, however, is an achievement that should not be overlooked. Certainly, it would have been seen as equally unlikely by the audience of our original talk, and the fact that the Ninety-Five Second Railway seems possible as well should not be overlooked. With that in mind, perhaps we should not bet against our visionary speaker’s dream coming true just yet.

Like what you read? You’ll find more in our magazine

Read in-depth articles about the past, present and future of transport in London and beyond. All in a beautifully laid out print magazine that you can read at home, work or on your commute. Buy it now

Written by Pedantic of Purley