Uncircling the Circle: Part 2


In part one of Uncircling the Circle we described how London Underground and its predecessors had tried, unsuccessfully, to eliminate the Circle Line as a continuous circle from the underground system.

We also looked at some very rational reasons why continuous circle operation is a bad idea. Despite that, there is of course always the nagging doubt that this is just something that London, or at least its transport providers, has developed a great irrational antipathy for. To investigate this we do something quite rare on London Reconnections and look abroad for a few examples of running lines in a circle – or not. Rather than consider all worldwide situations we confine ourselves to Western Europe where railway practice tends to be similar to that found in the UK.

Experiences Abroad

When comparing London’s Transport system it is almost inevitable that one looks at Paris first. A look at the Paris Metro map confirms that there are no circle lines. In fact most Metro (not RER) lines in Paris do not even have diverging branches and for one of the two that do, Ligne 7, there is a long-term proposal to extend ligne 14 to take over one of the branches. In other words, Paris likes simple metro lines and does not even like having junctions let alone circles.


The proposed Grand Paris Express Lines. Note how Ligne 15 has an extra couple of station on a protrusion added to avoid the line being a closed loop. Map courtesy of Wikipedia.

Possibly of interest is the Grand Paris Express plan for Paris. Of particular note is Ligne 15. It is planned to go around in a circle, but only once and it will have a branch off it to for a sort of lasso shape. The shape of the line suggests that a deliberate effort was made to avoid going around continuously in a circle.


Linie U3 of the Hamburger Hochbahn. It is remarkably similar to London’s current day Circle Line and is even portrayed in the same colour. Again the map is taken from Wikipedia.

Hamburger Hochbahn (Hamburg’s High-level Railway) line U3 used to run in an circle but no longer does so. Not only did it “uncircle” at the same time as London, but the solution employed was topologically identical.


Oslo’s Underground. Lines 4 + 6 are really just one line portrayed as two separate lines to avoid confusion about which “turn of the circle” it is. Yet again we have to thank Wikipedia for providing convenient access to the diagram.

In Oslo the Norwegians went to great trouble to complete a circle line called T-baneringen (the deep railway ring) – and then to even greater trouble to ensure that trains did not travel continuously round it. The result is a situation where you can leave the terminus on line 6, enter the loop and go round it and then find your train is announced and labelled as a line 4 train going to a different suburban terminus. The rough equivalent in London would be boarding a District Line at Wimbledon, doing one and a half loops of the circle and discovering that you were on a Hammersmith & City train destined for Barking. Conceptually this is similar to the idea of a “once round the loop” Hammersmith & City Line that was apparently dismissed by London Underground as too complicated for passengers.


Tyne and Wear Metro has two lines. Again it looks like it has been specifically designed to avoid the yellow line going round in a continuous loop.

Back in the UK, but outside London, we also have an obvious example of carefully avoiding running trains in a circle. In Newcastle the Tyne and Wear Metro Yellow Line actually crosses itself. Interestingly, the choice of colours suggest that they were chosen with thoughts of the London Underground in mind.

Of course much nearer home we have the London Overground – which deliberately doesn’t run through trains from Clapham Junction to Clapham Junction via Highbury and Islington. TfL and the Mayor might claim that they have an orbital railway but this is hyperbole. They have two lines from Clapham Junction to Highbury and Islington and by changing you can go round in a circle and get back to where you started. There is nothing special in this, and on the same basis one might as well claim that the Northern Line is an orbital railway.

The Exception Proves The Rule

There is a frequently misunderstood English expression: the exception proves the rule. A better phrasing would be the exceptional case proves (or disproves) the rule. In other words only by looking at exceptional cases can you tell if a particular rule is always always true.

An exceptional case for circles in the UK is is the Glasgow subway, colloquially known as “the Clockwork Orange”. Here there is really no alternative possible as it is a self-contained system built in a circle. Moreover, the trains running in a circle are the only trains running at all so there is no issue with them getting in the way of other trains.


The Berliner Ringbahn serves the four main suburban interchanges in Berlin. It runs in a circle and completes the circle in exactly one hour with the main four interchange stations approximately 15 minutes apart. Many thanks to Luciwest for the photo taken from her excellent one minute video of the Ringbahn which contrasts the very different areas that the line passes through.

A more interesting case is the Berlin Ringbahn which connects the four main suburban interchange stations as well as others in a continuous loop. It has long been a desire to run the service as a continuous loop as was done many years ago but this has only been possible since 2006 when more powerful rolling stock meant that the loop could be comfortably circled in just under an hour. Of course the timing of an hour isn’t absolutely critical but it does make it much easier to organise a regular interval service to match demand.

The Berlin Ringbahn operates every five minutes at peak times and ten minutes at other times. This very neatly gets over the problem of switching from off-peak to peak to off-peak again. The timings of the off-peak trains do not change and a peak-only train is inserted between two off-peak trains.

One may wonder why the Berliners run a train in a circle when almost no-one else does. Part of the answer will of course be that this is the best way to serve passengers and if they can run it reliably then why not do so? This probably isn’t the entire story though and virtually no decision is made in or about Berlin without reference to the city’s extraordinary history. The city had a circle line before the wall was built which resulted in it being severed. It is almost inevitable then that restoration of this service would be a highly symbolic act showing how Berlin had come together once more.

Perhaps it is not a case of never run trains in a circle just that you should only do it if you have a very good reason for doing so.

Back to London

The first commissioner of TfL, Bob Kiley, was an American who was adamant that when the SSL were modernised it would be done in one go, with one large rolling stock order to make it more attractive to suppliers. If one was to replace the rolling stock in one “hit,” it made a lot of sense to plan the upgrade holistically. It must have been quickly apparent that the Circle Line was once again a thorn in London Underground’s side. One can easily imagine a new-found determination to get rid of it in its current form and replace it with something more workable.

Apart from the many problems already mentioned, there must have been the additional major problem that London in particular suffered from. Either one builds the entire timetable around the Circle Line and accepts any consequential limitations on capacity or it becomes very difficult to fit other trains into a timetable.

On systems like the SSL with automation one could reasonably expect to run between 30 and 33tph. Because the District Line has diverging branches and the Hammersmith & City and District share a long length of track it makes a lot of sense to choose an even number for frequency if trying to run a regular service that repeats each hour. Choosing a number of trains per hour that divides by four is better still, so 32 is a very natural number to aim for. This then more or less forces a Circle Line service of 8tph but, as we have seen, with a trip around the circle taking slightly over 60 minutes this was going to be very problematic. Clearly the issue of getting rid of the continuous circle would not go away.

The T-Cup Service

As part of the staged plan for SSL, TfL included a proposal to finally reorganise Circle Line running. The first stage would see the Circle Line switch from running in a loop to one that, as today, started at Hammersmith and progressed to Edgware Road before completing one full “lap” round to Edgware Road again and then reversing. This was originally planned to happen around 2009 and was dependant on the service being provided by new six-car ‘S’ stock trains. It was suggested that the service diagram would look like the outline of a tea cup – presumably one with the bottom of the handle missing – and thus as the T-cup service it became known.

Apart from improving reliability on the Circle Line, the big advantage of the T-cup service would be that it would give a vastly overdue improved combined service on the Hammersmith branch. This had been an aspiration for some considerable while – not least because it was felt that the stations, which were all in zone 2, should have a comparable service with other zone 2 stations. This became even more of an issue with the opening of Westfield Shopping Centre. London Underground had opened a new station (Wood Lane) to serve it, but nothing had so far been done to improve the service to it.

2009 – Problems and time for a decision

By early 2009 London Underground had a major problem. More passengers meant longer dwell times at stations and Circle Line trains were now struggling to do a “lap” inside 60 minutes in peak times. By now the SSL upgrade should have progressed much further than it had, if not be finished, but the collapse of the PPP initiative put paid to that.


As the Circle Line used to be run with rush hour frequencies.
With 14 trains available this relies on the trains completing the circuit in exactly 60 minutes.

The 60 minute lap was critical because the Circle line was allocated ‘C’ stock trains – 7 for clockwise and 7 for anti-clockwise. To otherwise maintain the same level of service (7tph) one needed at least two extra trains but there were none available. Alternatively one could extend the time taken to go around the circle by 10 minutes (resulting in a 6tph service) but losing those 10 minutes was practically impossible without delaying other trains. Even a thoroughly unsatisfactory 6.5tph would mean completing a circle in exactly 65 minutes – probably not realistically achievable either.


How the peak service used to be. Almost the entire Circle Line had to work in conjunction with other lines and the frequencies involved allow no leeway in the timings or order of trains.

The Circle Line’s punctuality was by now consistently the worst of all the Underground lines. The knock-on effects were also dragging down the District and Metropolitan. This then potentially affected the Piccadilly, which shares sections of track with both the District and Metropolitan.

The obvious solution was thus to introduce the T-cup service much earlier than originally planned, at least in terms of the order of events constituting the upgrade. The first problem, however, was that the new stock wasn’t ready. The second was that even if it was ready, a decision had been made to go for seven carriages rather than the originally planned six – so there was also a lot of infrastructure work to do. Thus even if the ‘S7’ could be delivered in advance of the ‘S8,’ the required infrastructure changes along the Hammersmith branch and elsewhere had yet to be made. In any case, the need to replace the A60 stock on the Metropolitan Line with the S8 was if anything more urgent.

Making the T

Because there was no new stock available it was not possible to run 7tph on the T-cup service. However it was possible to run 6tph on the three main ‘C’ stock routes combined – Circle (if now a T-cup service), Hammersmith & City and Wimbledon-Edgware Road (“Wimbleware”) services. This could be achieved by a marginal reduction of frequency (except on the Hammersmith branch which would have more trains) and abandoning an out-of-pattern single ‘C’ stock service that ran in the peak hour between Wimbledon and High St Kensington.

The only problem with introducing the T-cup service early with this reduced frequency was that in the peak hours other services would have to be awkwardly adjusted to fit in. The Wimblewares were not a problem north of Earl’s Court, since they would also be 6tph and the western part of the Circle Line is not especially busy. The Metropolitan from Baker Street to Aldgate was 14tph which neatly fitted in with 7tph on the Circle and 7tph on the Hammersmith & City. Surprisingly, in the revised timetable this went up to 15tph – presumably to partially compensate for the loss of 1tph on the Circle Line and 1tph on the Hammersmith & City Line. The Metropolitan overall had a small increase of 0.5tph to go to 22.5tph which was presumably made in the belief that arrival of the first ‘S’ stock was imminent and a further timetable change was to be avoided if possible.

13th December 2009 – T-Day

In March 2009 TfL announced that the new Circle Line arrangement would start in December of that year. London Underground was pretty much committed to permanently running a revised Circle Line despite never having tried it out. The impression given was that they believed things just couldn’t be worse than they already were. The unanswered questions, however, were whether it would actually work and, perhaps more importantly, how it would it go down with the travelling public.


The previous off-peak frequencies. This meant 21tph (same as in the peak) through Praed Street Junction.

The main fear, operationally speaking, was whether Praed Street Junction – controlled by Edgware Road signal box – could cope with the additional trains. At first sight the increase appeared to be considerable as the junction now effectively had four lines running through it compared to three previously. In fact, due to a reduction of frequencies, there was only an additional 3tph in each direction (an increase from 21tph to 24tph). This was for the best, as Edgware Road is a very old signal box, and it would not have been realistic to make major alterations at this stage.

Current Service C stock

The service today both peak and off-peak. The number of trains each hour per direction through Praed Street Junction has only gone up by 3 to 24tph. The number in red in brackets shows the number of trains terminating.

To reassure themselves that the T-cup service was workable, in the summer of 2009 London Underground decided to introduce it just for some weekends, when engineering works meant an alternative service had to be arranged in any case. Although from a passenger’s perspective things did not appear to go too well, with both staff and passengers confused as the nature of the service, the trials must have been judged an operational success because London Underground pressed forward with their already-announced plans for a permanent December implementation.

One feature of the implementation that surprised many that December was the sense of permanency about the change. A major operation was carried out to replace signage so that passengers would be less confused than they would otherwise be. For some items such as posters of the Underground Map this was fairly easy to do, but changing the many enamel signs that referred to the Circle Line was not and it was probably this major effort that convinced people that this was not a speculative change.

Inevitably, as always with changes to people’s travelling habits, there were teething troubles and the usual stories of doom from the Evening Standard. Reliability, however, improved both swiftly and dramatically and for the most part people adapted quite well and quite quickly. The main benefit appeared not to be that the revised arrangement prevented delays, but that delays (however caused) on any of the SSL could be recovered from much more quickly. Even the Evening Standard realised this major change was rapidly becoming a non-story.

The Paddington Issue

The main problem, one that remains today, is Paddington. The issues there are quite involved so we are actually going to cover them, as well as provide a photographic update on the station, in a later article. Suffice to say for the moment, a change in passenger habits was required for those arriving at Paddington and wishing to continue onward eastbound to the northern part of the Circle. They would now be better off using Paddington (Hammersmith & City) station and thus avoid needing to change trains at Edgware Road.

For the passengers that are forced to change at Edgware Road there was at least the mitigating factor that in a clockwise direction there was a 50-50 chance that it would be a simple cross-platform interchange. In the anti-clockwise direction there would also be a 50-50 chance that the first train would be a simple cross-platform change. If it wasn’t, then the options were to use the stairs to change platforms or to wait for the second train. Whilst the C stock remained the wait could be in a cold and exposed train with the doors open. When the C stock has been replaced by the new air-conditioned S stock this unwelcome experience should be alleviated by a feature whereby the doors close themselves at the terminus after being open for 45 seconds. At least the wait should be more pleasantly warm in winter and cool in summer.

Edgware Road Signal Box

As indicated earlier, a critical factor in the success of the T-cup service was the ability of Edgware Road signal box to handle the additional trains. Its age has prompted concern from people outside the rail industry, notably Assembly Member Murad Qureshi, but as stated by the Mayor in a written answer to Murad in Mayor’s question time:

the upgrade of the entire sub-surface railway signalling system (including the signal cabin at Edgware Road) is expected to be complete in 2018. It would not be practicable or effective to attempt to replace the signal cabin in isolation from the wider signalling upgrade. The signalling cabin, whilst old, continues to be maintained and to operate safely.

In fact it would have been more sensible if Mr Qureshi had asked if there are any sites more critical than Edgware Road – to which the honest answer would undoubtedly have been “Yes”. There is a great irony in the state of old railway equipment in that the very old equipment is often easier to repair and maintain than some of the old but newer equipment. This particularly applies to very old electronic equipment and it is probably just as well Mr Qureshi was not shown the old (but technically much newer) electronic control system that the Bakerloo Line currently relies on.

The signal cabin in question was in fact updated in 2011 but only for the purpose of altering the track circuits to make the track layout at Edgware Road station suitable for the longer S7 trains. Rather curiously the press release stated that Praed St junction is “the busiest junction on the Tube network”. In reality it is not even the busiest junction on the Circle Line and probably isn’t even in the top three, which shows how much caution one perhaps needs to apply to press releases. In fact its notoriety comes from the age of the equipment and not the frequency of trains passing through it.

It seems then that by and large the idea of reinstating the Circle Line as it was is now largely forgotten. It does occasionally surface. For instance in December 2012 the Guardian took up the cause. One would be more convinced by their arguments if the did not bring up the red herring of Paddington stating:

So, passengers have to change at Edgware Road, often dragging luggage they may have brought from Heathrow via the Heathrow Express up and over a Victorian footbridge to await, with negligible public information, another train. History and passengers should demand the return of the Circle line.

As we saw earlier, as far as passengers arriving at Paddington are concerned, this is incorrect. Direct trains are available (and frequent) if you go to the correct station. With tongues at least partially in cheek it is tempting to wonder whether any of the now-King’s-Cross-based paper’s journalists reside in Bayswater or Notting Hill – two of the few areas adversely affected by the change.

Murad Qureshi picked up on the Guardian report and joined in the call for the Circle Line to be reinstated. As earlier, he seems to be convinced that the problem lies with Edgware Road signal box and says “Improve this and you could have more trains going through the junction in both directions”. Of course this is probably true, but as always with these sort of things one has to look at the larger picture. It is difficult to see how one could take advantage of an increase in the hypothetical capacity at Praed St Junction as one would undoubtedly be constrained elsewhere.

There are no doubt some people that have lost out from the revised Circle Line. Many though – especially those using the Hammersmith branch west of Paddington – have gained considerably. The losers seem to be confined to those from High St Kensington, Notting Hill Gate and Bayswater stations who wish to travel beyond Edgware Road. So the problem, if there is a problem, would be to find a solution to rectify this. One possible solution is to certainly to bring back the Circle Line as it was but another option would likely have to be found. Bringing back the previous pattern would now have enormous ramifications affecting everywhere from Richmond to Upminster. It would also have knock-on effects that would affect reliability everywhere from Chesham to Wimbledon.

It does seem though most of London accepts that the Circle Line as it was will never return and no doubt some of those would not wish to see that happen anyway. As it is, with the inherent complexity of running the Sub-Surface, it is hard to see significant deviations from the established plan taking place before the 2020s.

For now at least, the Circle has been – and will remain – uncircled.

Written by Pedantic of Purley