Little Things That Matter: Railroad Junctions

One underrated difference between countries is how multi-tracked railroad junctions look. In France, double-tracked regional lines have grade-separated junctions that ensure no crossing oncoming traffic. For a plethora of examples, consult the RER track map and look at any bifurcation. Looking at Google Earth, the same is true near Tokyo. This is standard rapid transit practice anywhere I know of, and Paris and Tokyo both treat their regional rail systems like urban rapid transit.

In the US, this is not true. Even important, high-traffic mainline junctions are often flat – see for examples the Main Line-Hempstead Line junction on the LIRR (Queens Interlocking), and the Hudson-Harlem junction on Metro-North (Mo Interlocking). The major junctions involving the Northeast Corridor tend to be better, fortunately. Harold, the LIRR/NEC junction, is already grade-separated from oncoming traffic, and the current grade-separation project is only for same-direction traffic; and the junctions in New Jersey are grade-separated. The Kearny Connection splits the problem in half – it is grade-separated for NEC trains but requires Morris and Essex trains in opposite directions to cross each other at grade. However, even for NEC trains a few major problems remain, most notably Shell Interlocking between the Northeast Corridor and Metro-North in New Rochelle.

I suspect the problem is that double-tracked lines in the US are not consistently thought of as having one line in each direction. The arrival of centralized traffic control (CTC) has made wrong-direction running easy; some railroads ripped their second tracks, and the commuter lines that remained double-tracked freely run trains wrong-way during weekends or (as is the case on the Worcester Line) when there are freight trains on the line. At a few places, four-tracked segments on running track connect to two tracks in nonstandard ways: for example, at Providence Station, three of the four platform tracks merge into the southbound running track. The concept of having one track per direction and no crossing oncoming traffic, which is standard on the subway, doesn’t really apply to commuter rail, leading to scheduling problems.

In New York, there’s no alternative to grade-separating the worst junctions, including Mo, Queens, the Kearny Connection, and the unnamed Far Rockaway/Long Beach and Ronkonkoma/Port Jefferson junctions. Although frequent train service exists with flat junctions, the schedule is irregular and unreliable, and has few reverse-peak trains. Fortunately, this is a problem for commuter trains more than for intercity trains, for which schedule adherence is more important.

In Boston, the NEC itself has flat junctions at all of its branches. Fortunately, there are alternatives to concrete. The Franklin/Providence junction requires Franklin Line trains merging onto the NEC to cross oncoming Providence Line trains at grade, but lets them continue onto the Fairmount Line without conflict. Since the Fairmount Line is getting some investment and more frequency is under discussion, having additional trains serve the line is a net benefit, and all Franklin Line trains should go through Fairmount. The Needham Line branches at-grade, at a more constrained location, but there are plans to connect it to the Orange Line anyway, and much of its geography is suitable for subway service more than for a regional rail branch. This leaves the Stoughton Line, for which there’s no alternative, but fortunately Canton Junction is not a very constrained location and the junction is simple.


  1. Tom West

    The UK rarely grade-seperates rail junctions. However, even where the signalling allows bi-directional running, you’ll find a given track had almost all trains running the same way. This is true even at termininus stations… for example, at Euston, there are six tracks approaching the station, all bi-directional from a signalling point of view. However, they are almost always used in three pairs of up and down.

  2. Aaron M. Renn

    Reminds me of Metra’s A-2 interlocking, where the three track Milwaukee District line from Union Station crosses the four track UP-West line out of Northwestern Station at grade in close proximity to the Western and California coach yards. The need for this crossing could probably be eliminated entirely by simply switching the downtown terminals of both lines. (There’s already integrated fares and fare media). But that would require organizational work, which no one wants to do.

      • simple

        Also much easier said than done. For one, it would involve a billion dollar demolition (and future rebuild) of the office building currently above the commuter concourse. Then, once you’ve done that, what would your new CUS track and platform layout look like? And how exactly would you use all of those through tracks? I’m not saying there may not be solutions out there (yet there indeed may not be – the track geometry is very tight, and simply replacing 20 stub-end platforms with 10 uselessly long through platforms is not going to increase capacity), but certainly none as easy as such a casual comment might imply…

        • Beta Magellan

          I’ve read $700 million for tear-down-rebuild-replace as a rough estimate,* with the figure coming from someone associated with MWRRI’s informed fantasy planning.

          While it would work fine for UP-W, you would probably still need Ogilvie/Northwestern station for UP-N and -NW, unless you’re willing to close Kinzie Street to those lines down to the level of the Milwaukee District tracks. That’s probably why when CDOT talks about through-running they mention a new tunnel under utility-free Clinton, with inclines for the N and NW lines between Kinzie and Grand and another exiting at Desplaines for westbound lines—it results in a less complicated, crossing-free approach for all the lines, and since all trains through the tunnel would run-through terminal capacity’s not a problem.


    • simple

      Much easier said than done. Right now the north side of CUS sees about 157 trains per day (in+out, but not counting yard deadheads, which nearly doubles that figure) and Ogilvie 153. Under your plan CUS-N would see 59 and Ogilvie 291 (or if we made Amtrak an exception to your rule it would be CUS-N 75 and Ogilvie 275). Assuming peak-to-base is roughly the same on all of these lines, that means nearly doubling the peak hour train movements at Ogilvie. I agree with Alon’s frequent claims that fast(er) turnarounds of commuter trains are technically possible, but is it really just an organizational matter in this case? Hardly.

      • Nathanael

        I proposed a complete swap of MD and UP lines; there’s ROW to get the UP N and NW lines almost to the Milwaukee District tracks (one smallish building is in the way). Amtrak has to stay in Union. How do the numbers work out then?…

        The organization problems are *serious*, however. Union Pacific operates the UP lines, but not the others. It is also notoriously hostile to passenger traffic, and to change, and would probably be unwilling to do a swap.

    • Beta Magellan

      The big functional issue with the proposal comes not at the terminals (as simple implies—if we look at peak hour schedules and only assume Ogilvie can use its four full-length platforms, it comes out to 7-8 trains per hour per platform, which is perfectly workable—there are also four shorter platforms to work with too), but at the interlocking itself—trains can only cross between UP to Milwaukee District tracks at fairly low speeds, which basically undoes any of the advantages of the getting rid of the crossings (I’d suspect it might even make things worse).

      Metra is planning to eventually get rid of the A-2 crossing—they’re planning on making a new, more modern crossing at Ashland (called A-1). Although this would improve operating speeds (and therefore capacity), unfortunately it doesn’t look like they’ve included any provision for peak-period switching between the lines or grade-separating. I doubt this is a Bay Area-style bid to eventually build a megacomplex—rather, it simply strikes me as lacking imagination.

      • Nathanael

        The proposal would be to straight-line the tracks with no crossing at A2, which could be done relatively easily, really.

  3. Steve

    Part of the reason that the NEC in NYC and Philly has so many grade-separated interlockings is because of the legacy of the Pennsylvania Railroad (PRR), which, at one point some ninety years ago, was considered to have the most advanced trackwork and interlocking engineering anywhere.

    Unless I miss my guess, Shell was once the PRR-New Haven (R.R.) junction; today’s NEC north of there follows the former Along the Shore Line.

    Addressing interlocking efficiency really needs to be a goal of established commuter rail lines, however. The primary goal for most new commuter rail is to get it going, and if that means tolerating minor inefficiencies (and I’m sure you’ll agree that in most situations interlocking throughput efficiency is minor), then so be it. Costs do need to be the dominant arbiter in the initial startup phase; greater interlocking efficiency should be a byproduct of throughput demands encourage (or even necessitating) it.

    • anonymouse

      Shell was strictly New Haven territory and was originally built as the junction where the mainline split into the passenger line toward Grand Central and the primarily freight line to the Harlem River terminal, which eventually was also connected to the NY Connecting Railroad (jointly owned by the PRR and New Haven) to Bay Ridge for the car floats and Penn Station for through passenger service.

      As far as junctions on the Boston section go, the flat junctions with the Needham and Franklin line are somewhat mitigated by the ability to use Track 3 in either direction. This means that in the morning, you can use it for inbound Needham and Franklin trains, leave the usual northbound track (Track 2) for inbound Providence/Stoughton and Amtrak trains, and the southbound track for all southbound trains, leaving no conflicts at all until you get to South Station. In the afternoon, Track 3 can be used by outbound Franklin/Needham and some Providence trains, and northbound (reverse peak) trains end up having to cross, and conflict with, everything, but there’s not that much reverse peak service (or demand for it), and there’s probably enough capacity to deal with it. Likewise, Canton Junction is a problem, but likely not the biggest constraint on capacity. Right now, for commuter rail at least, I think it’s the fact that Ruggles and Back Bay are in the same signal block, which dramatically reduces capacity. The lack of a Ruggles platform on Track 2 also introduces all kinds of annoying constraints into the timetable that add all sorts of crossing conflicts. That needs to be addressed first, before any of these junctions become the main constraint. In the long term, converting Needham to an Orange Line extension and sending more trains via Fairmount ought to leave more than enough room for the 4 tph of Providence service, 2 tph of Stoughton service and whatever Amtrak and regional trains need to use this line.

  4. lpetrich

    In the Chicago area, there’s a massive set of grade-separation construction projects called CREATE: both road-rail and rail-rail. These projects are in various stages of completion, from planning to under construction to operational.

    In Los Angeles, there’s now the Alameda Corridor, a grade separation for a 20-mi / 32-km route between downtown LA and Long Beach. It is now being extended eastward from downtown LA.

    The San Francisco Bay Area has several commuter/intercity road-rail grade separations, like for Caltrain, but hardly any commuter/intercity rail-rail ones. BART is entirely grade separated, however. It has four junctions, downtown Oakland, MacArthur, Bay Fair, and SFO, with complete rail-rail grade separations. But BART falls in the tradition of urban heavy-rail “metro” systems.

    • simple

      The seven CREATE “flyover” projects all involve grade-separating railroad grade crossings (diamond crossings) — not grade separating railroad junctions. Operationally, this is a very different thing from what Alon is describing. In the case of CREATE, the flyovers are principally to keep passenger/commuter trains and freight trains off of each other’s paths, so frequent passenger/commuter service during peak periods doesn’t effectively create a 3-hour shutdown every morning and every afternoon for the affected freight lines.

    • anonymouse

      The SFO junction is not fully grade separated if used as a stop along the way to Millbrae. If two trains arrive from SF and Millbrae at SFO at the same time and depart to Millbrae and SF respectively, there’s going to be a crossing conflict either coming or going. But yes, that’s the only place where that can happen on the BART system. For Caltrain, the only real problem is conflicts at the terminals, particularly with Amtrak or ACE trains departing from Track 4 at San Jose.

      LA has a lot more problems in this respect: there’s Burbank Junction where the Coast Line diverges from the Metrolink Antelope Valley Line, the junctions where Metrolink/Amtrak join the BNSF main (Fullerton Junction and Hobart), the junction between the Riverside Line (UP LA Sub) and the BNSF main at Riverside, and either end of the line from Orange toward Corona. And of course the big one is the Colton Crossing, which is actually going to get fixed soon.

  5. Adirondacker12800

    Even important, high-traffic mainline junctions are often flat – see for examples the Main Line-Hempstead Line junction on the LIRR (Queens Interlocking)

    Except that the Queens Interlocking is somewhat east of there. ( assuming lirrtrackmapv3 from R.E.G. is correct – it seems to be from the satellite images of Bellerose ) The Main Line and the Hempstead Branch disentangle themselves east of the Queens Village Station. At Floral Park the trains on the northern pair of tracks are Main Line trains and the trains on the southern tracks are Hempstead Branch trains. A shorter version of what happens between Jamaica and Valley Stream.

  6. Adirondacker12800

    and the Hudson-Harlem junction on Metro-North (Mo Interlocking).

    And those wily devils at Metro North seem to be separating at least some of the Hudson line trains from the Harlem and New Haven trains.
    Starting at 0:25 ish

      • Alon Levy

        Yeah, I see it. It’s workable on a railroad with a fair amount of unused track capacity, and an enormous CBD train parking lot. It wouldn’t work as well if it weren’t for the 3-and-1 running at rush hour. The alternative would be to make the western two tracks permanently Hudson Line-only and the eastern two tracks Harlem and New Haven-only, but this is impossible given the total demand on the Harlem and New Haven Lines. Even diverting New Haven trains to Penn Station would barely allow it, and only if there were no increase in Harlem Line traffic.

        • Adirondacker12800

          So they should pour concrete in the Bronx even though by leveraging the signal system’s capabilities and a bit of thought they manage to do it without pouring concrete?

    • Walter

      Yea, the Hudson Line are usually segregated to the west side of GCT, and use the track 4 platform (westernmost) platform at 125th Street before diverting at MO. Usually the only time a Harlem Line train will be over there is if they need a diesel or if they are short equipment for a White Plains local. I’ve only ever seen a New Haven Line M2/4/6 consist on that side once.

  7. Richard Mlynarik

    Alon, bi-directional signalling has pretty much nothing to do with it.

    It’s all about traffic density (outside the NY metro area there pretty much isn’t any on US RRs) and schedule adherence (an utterly foreign concept.)

    And capacity through junctions is far from a “little thing”, unless there are so few trains that you should be running buses.

    Once traffic levels are non-trivial (more than ~4tph at decent speed), flat junctions become scheduling problems and wrong-track running (which if you think about it is a what flat junctions involve) go out the window, except for the shortest sections approaching the flat throats of terminal stations.

    To get on my usual fan hobbyhorse, there are some pretty spectacularly impressive examples of flat junctions in (where else?) Switzerland, but these require operational heroics and they’re being dealt with just as fast as capital priorities demand. “Elektronik vor Betlon” goes a long, long way, but eventually something has to give as train density goes up and up. (Lausanne-Renens, Hürlistein, Liestal Nord (approaching Basel), Olten Nord, Olten Ost, Basel Ost maybe, Pfäffikon maybe, Dorfnest maybe, etc, etc, … the wish list is long, so strategic prioritisation is necessary.) The Netherlands has many great (ie “bad”) examples also.

    PS You’re sure to like this: The Funnel into Zürich (in 2001)

    • Alon Levy

      The bidirectional signaling bit is why several railroads single-tracked their lines. And more in general, it’s causing railroads (both freight and passenger) to think of double-tracked lines as two adjacent single-track lines sometimes…

      Also, I know that junctions aren’t a little thing. But they’re unsexy, and do not figure as prominently in plans for modernization, whereas grade separations, even road-rail separations, do. If e.g. the MBTA wants to pull itself out of the 1930s, it will need to deal with them (as well as other unsexy things, like the aforementioned issue with Back Bay and Ruggles being in the same signal block). It’ll improve quality of service if they electrify and run modern rolling stock frequently no matter what, but a lot of the improvement will go down the drain if because of the junctions the scheduling is too weird.

      • Richard Mlynarik

        Sure Alon. I thought you were making the argument somehow that a US culture of bidirectional signalling made flat junctions less troublesome than in some Euroland or Asiaworld technical incapable of wrong way running; the reality being that it the same basic factor — traffic density — that wants to see single-direction tracks and conflict-free routes.

        PS Here’s some classic “Elektronik vor Beton” action for you. This scan (from the must-own book ISBN 3909111068 (German) 3909111076 (French), sorry I only have it in DE) explains how a flying junction was temporarily avoided at one of the national network’s super-critical locations west of Olten by building adding a short section of fourth track and providing signals and crossovers and software to allow trains to cross each other at a flat junction at one of two different locations: one for regular exactly on-time operation, and a backup that allows the conflicting train to be delayed two minutes without causing knock-on delays to the other. This stunt is still performed twice an hour, every hour. The deferred flyover will be built eventually, but putting that expense off for 20 years and allocating resources elsewhere is nothing to sneeze at. Work smarter, not harder, and all that.

        • Andre Lot

          This looks just like a cheap (in the pejorative sense) solution. It is like devising a super-neat protocol to put some buckets on the floor when water is pouring in instead of fixing the ceiling pipe leak.

          Moreover, it is only viable because of anti-competitive practices that give the railway total control over circulation of trains. As one who thinks competition (true one) is a paramount value that supersedes all other concerns in railway operations, I wish this kind of solutions disappears, as they oblige schedules to be too much coordinated, reducing the ability of non-cooperative, competing operators to run trains in uncoordinated schedules in the same area.

          • Alon Levy

            SBB has some nasty things to say about putting competition above everything; it’s rejected the EU solution of keeping tracks and operations separate, on the grounds that what Switzerland has right now works fine.

            But even outside Switzerland, this could work. Germany has integrated scheduling with operations put out to tender; its train system is less tightly integrated than Switzerland’s, but that isn’t really because of open access. And Japanese railroads run trains punctually on their own systems; the competition there is between integrated operations-infrastructure(-development) corporations, rather than between franchisees as in Britain and soon the rest of the EU. There are no flat junctions in Japanese urban areas because the traffic level is far too high for that, but there do occur other instances of running trains on a schedule on less than perfect infrastructure – for example, the outer end of the Yokosuka Line is single-tracked with passing sidings, and has 15-minute service.

          • ComradeFrana

            “As one who thinks competition (true one) is a paramount value that supersedes all other concerns in railway operations,…”

            Well, I am pretty sure that people in SBB think the paramount value that supersedes all other concerns is the value that the railway system has for the end user. I think they are pretty successful at that.

            “…as they oblige schedules to be too much coordinated…”

            That’s not a bug, that’s WAD. And that’s an understatement, they are planning multi-decade infrastructure improvements just so their schedules can be even more “too much coordinated” to maximize connectivity and system legibility.

            “…reducing the ability of non-cooperative, competing operators to run trains in uncoordinated schedules in the same area”

            No, what reduces the ability to run trains on uncoordinated schedules the most is the sheer frickin’ volume of trains there is. And on the contrary, competent train coordination and dispatching (such as the swiss one) actually helps trains with variable schedules, namely freight, fit in the limited windows between regular trains.

            /rant over

          • Andrew in Ezo

            Alon, there are a few flat junctions left in Japanese urban railway settings, albeit they are being grade separated as we speak. The most famous is the Keio Railway junction at Chofu:

            The Osaka representative, at Hankyu Railway’s Awaji Station:

          • Nathanael

            “Moreover, it is only viable because of anti-competitive practices that give the railway total control over circulation of trains. As one who thinks competition (true one) is a paramount value that supersedes all other concerns in railway operations,”

            I hope you reconsider this ideology.

            Railroads work *extremely well* when *highly centralized*. Competition is usually neither helpful nor valuable for them.

            This is, of course, why they need to be democratically controlled, and the Swiss railways seem to, on the whole, be controlled quite directly by the legislature, or sometimes directly by the referendum system.

            It only works in a country where average people recognize that railroads are valuable, though. Sigh.

  8. simple

    Also it’s worth noting that flat junctions are not entirely absent from rapid transit railroads — although they’re indeed not standard design anymore. But CTA’ L and London’s Tube both still have several outrageously busy ones, thanks to penny-pinching by the original railroad builders more than 100 yrs ago. Tower 18 on CTA sees over 1,000 trains per day and Tower 12 and Clark Jct both over 800. Of course this leads to lots of delay, but its tolerable as it’s concentrated over a small section of tracks where trains are going slowly anyway (somewhat akin to the throat of a terminal). Needless to say, whenever there’s a problem with one of these junctions during a peak period, it’s a complete meltdown for the commute!

    • Nathanael

      I’m not sure it quite qualifies as penny-pinching in either of those cases. In Chicago, the Loop was built in the restricted space over the downtown streets, surrounded by tall buildings, and a flying grade separation would probably have been unimaginably expensive. (Assembling the property and permissions for the Loop was already a feat, with individual companies set up for single blocks.) Of course, the same situation meant that there are sharp 90 degree curves everywhere and all the trains are going really slowly anyway, so the flat junctions are hardly the worst problem. (Clark Junction is less excusable.)

      In the case of London, the flat junctions are in the cut-and-cover lines, again built underneath major streets which were already surrounded by tall buildings. Building a flyover there would have required a double-depth cut-and-cover construction and bridges inside the tunnels, and again probably was exorbitantly expensive. In this case, switching to deep bore tubes made all the difference, and all the bored tunnels have flying junctions.

  9. Beta Magellan

    Alon, from what I understand the Orange Line extensions have been quite dead for a decade or two. The MBTA didn’t see traffic from Needham as being heavy enough to justify new grade separations (which would also increase local resistance to the line) so they planned on having the Orange Line switch to overhead power at some point beyond Forest Hills and run at-grade with crossings (like the CTA’s Yellow Line did in Skokie) to Needham Heights. Although some of the Orange Line cars were outfitted with pantograph mounts, I’m pretty sure those have been removed and any new cars likely won’t have them either (and third-rail at grade is evidently unacceptable, even though I’m pretty sure more people are killed by collisions than electrocutions with Chicago’s surface-running rapid transit).

    However, a Needham branch for the Green Line was proposed by Boston’s MPO several years ago, but it didn’t make the final list of recommended projects. That document also has a rather grotesque Silver Line extension which follows the route of a mid-sixties proposal to deal with the Green Line’s awful E-Branch level junction west of Copley—a second subway under Stuart connecting Prudential with the unused Pleasant Street incline south of Boylston.

    • Alon Levy

      That list of projects just makes me roll my eyes. The worst one on the list is frequent service on the Fairmount Line, not to South Station, which is right on the border of the CBD, but via a new track connection toward Back Bay, for maximum capital cost and minimum utility. A couple of the infill commuter rail projects might be nice, but their projected costs are so out of sync with what’s normal ($10 million per new station, etc.) that they’re not worth it.

      On second thought, the Needham Line could be split between the Green and Orange Lines. Needham is quite suburban, so the service level of one branch of the Green Line could be adequate. But the parts of the Needham Line that are within Boston look urban on Google Earth and could use more frequent service, to a greater variety of destinations than Back Bay and South Station. The issue is that the cost of a connection to the Orange Line is about the same as that of a grade-separated junction connecting to the NEC, and the distance is short enough that the higher potential speed of good commuter rail would not be enough to make up for reduction in frequency and the need for an extra transfer to go north of South Station.

      • Matthew

        Roslindale certainly used to have streetcars, the old substation is still there and they’re trying to decide what to do with it.

        My impression is that people who live there tend to use the buses to Forest Hills in lieu of the Commuter Rail due to lack of frequency and high cost ($4.75 vs $1.70 after transfer to Orange Line).

        Rozzie/West Roxbury could definitely benefit from an Orange Line extension, but it’s not discussed much. May just be politics.

      • Beta Magellan

        I like an Orange-Green replacement for Needham as well (since the grade crossings happen in Needham, extending the Orange Line to West Roxbury should be very cheap). You made a great point on accessing points north of south station—West Roxbury’s home to a lot of civil servants, so service right to State St. and Haymarket would be ideal. I’m guessing the main reasons it isn’t discussed are political and institutional. From what I understand, the current Boston political establishment is fairly anti-transit—I get the impression that the MBTA gets more pro-transit pressure from Brookline, Cambridge and Somerville than Boston itself. Based on what I’ve read and a talk I attended given by a former MBTA higher-up, the T has come to see its main role in providing urban services as managing decline, not expanding the existing ridership base. Furthermore, I’m not sure how much those in charge actually know about transit, or at least rail transit—the former official said that the Green Line extensions into Somerville were impossible because they’d put the existing downtown tunnels over capacity—i. e. he didn’t know how through-running works.

        On the other hand, the MBTA is fairly fond of park-and-ride based suburban service expansion. I’d guess one reason why the document I linked to above only proposed rapid transit for Needham Heights, Center and Junction was because they wanted to keep the tracks east of Needham Junction available for a future commuter rail extension to Dover, Medfield and Millis (or rather large park-and-rides outside of each of those towns). Although this makes those of us concerned with frequent service to dense neighborhoods retch, it makes sense from the MBTA’s perspective—it’s a surer way to get federal money (lots of new transit riders from those park-and-rides—pity they won’t use transit for anything but commutes) and expand the number of legislative districts they run through.

        • Alon Levy

          The annoying thing about this approach is that even the MBTA’s stated approaches don’t work.

          First, there’s no decline to manage. Well, maybe the bus ridership is declining, but the Orange Line is holding steady, the Green and Blue Lines are slightly up over the last 20 years, and the Red Line is up nearly 50% (link). Although some parts of the system are in decline, e.g. Braintree, overall rail ridership is up, just not as much as in New York. Similarly, there’s no population decline to speak of. If one goes by ACS numbers rather than by census numbers, then Boston’s population went up 9.5% between 2000 and 2009 and Cambridge’s went up 7.3% (but Somerville’s went down 1.3%), vs. metro area growth of 4.5%. Even by the census numbers, with their urban undercount, Boston outgrew its suburbs, though by much less, and Cambridge was basically a wash with the rest of the metro area.

          And second, the MBTA isn’t doing well with getting commuters from the suburbs to Boston and back. If you compare suburb-to-Manhattan commuter flows, pretty much all commuters are accounted for by commuter rail, except in the GWB catchment basin, where the commuter rail sucks, and there many people take buses instead of driving. In Boston, no such thing holds. There are about 200,000 people commuting to Boston from the suburbs, not counting those served by urban rail, like Medford and Brookline. The MBTA has 70,000 inbound weekday commuters. (Comparing Boston to New York writ large is inappropriate, since New York contains most of its bedroom communities, whereas in Boston, a huge fraction of what would be outer boroughs in New York is outside city limits.) I’ve only eyeballed the commuter flows, but it looks like there’s a big surfeit of people north of Boston who are driving in, thanks to commuter trains that are slow and infrequent and dump them outside the CBD, and scary amounts of investment in freeways instead. The lines feeding into South Station are somewhat better, but still get less than 50% mode share. Commuter flow from Rhode Island to Boston is about 5,000; ridership at Providence is 2,000.

          • Matthew

            I suspect that many people on the north side are driving to Oak Grove or Alewife. Also, north side lines skip or don’t go to many of the more densely populated suburbs (Somerville being the poster child). Boston’s shape and transit are both weird. The city includes many areas which once were and still are – for all purposes – suburban. But the borders exclude many places which are more “urban” and transit friendly. Many of the neighborhoods inside the city are not served well at all, and many neighborhoods outside are served well. So commuter rail is not the whole story, buses are serving a large proportion. I do agree that the Commonwealth has spent disgusting amounts of money on highways and neglected transit, though.

            You’re right about North station but — try standing there at 5pm on a weekday. You will get run over by people rushing between the subway stop and the terminal. This may be mostly due to the fact that the transfer is a huge pain, requiring lots of walking up and down stairs, and via narrow corridors. Got to wonder what they were smoking when they designed it.

          • Alon Levy

            Possibly when they built North Station the Boston CBD had not yet fully formed, so the station’s location seemed reasonable. Certainly when they built the individual stations that were merged into North Union Station this wasn’t an issue – the small, spiky CBD is a modern development, and didn’t exist when the Boston and Lowell was being chartered in the 1830s. Why the more recent rebuilds have not included better pedestrian circulation is another matter. I’ll eventually do another little things that matter post on the subject; suffice is to say, American passenger train projects go for the flashy (new train stations, higher train speed) and neglect access and egress times.

            You’re right about Boston’s borders, and yes, ideally I’d also find a way to shoehorn Cambridge into the mix. (Cambridge might actually have a higher ratio of jobs to employed residents than Boston.) Somerville is a bedroom community, though; from having been there briefly twice, it looks like a smaller version of Eastern Queens. I went for quick and dirty, and am going to do a real post about this soon.

          • Adirondacker12800

            All trolley burbs look like Eastern Queens, the only thing that varies is the style of ColoniaL Revival they are slapping on it. If you hit one that’s mid or late Victorian you’ll find Victorian or one that is later and you’ll find Craftsman. It’s like tree rings in Western Queens, Brooklyn and the Bronx. Older styles closer to the stations and then newer styles farther out. In someplaces – where the railroad came through and then the trolley or subway arrived – you can see two or three layers of redevelopment.

          • David Alexander

            All trolley burbs look like Eastern Queens, the only thing that varies is the style of ColoniaL Revival they are slapping on it.

            It’s interesting that you noted that as I grew up in a southeastern Queens community that was on the fall line where the Colonials* would disappear and the cape cods and ranches would begin. I tend to end up using Eastern Queens as the yard stick to measure what I would call small density versus low density in which there’s single family housing and some mixed used development that tends to be dense enough to justify some degree of bus usage. While my trip to San Francisco have left me using denser parts of Brooklyn as a yardstick to compare their urban patterns, my trip to the Pacific Northwest lead me to use Queens as the yardstick. FWIW, IIRC, other than Great Neck and Floral Park, the density levels end up sinking a bit once you cross the county line, but compared to some modern Sunbelt suburbs and exurbs, the Town of Hempstead is transit friendly and “dense” as there are some areas within walking distance of a bus line** or commercial districts and small strip malls. Regardless, when I drive down Hempstead Turnpike in East Meadow & Levittown, there are times where I wish that they’d re-align the zoning to allow for some mixed use development and condos that aren’t restricted to seniors, but people fear their neighbourhoods turning into Queens…

            *And, yes, after twenty years, I learned to hate Colonials and envy the cape cods.
            **That doesn’t mean the bus service is usable. I’m within a 2 minute walk to a bus stop, but the bus comes once an hour until 7 PM with no weekend service…

          • Adirondacker12800

            For what it’s worth Queens is denser than San Francisco. There’s a glitch in the Census Bureau’s QuickFacts pages, the density numbers only go up to 9999.99. If I calculated it correctly Queens has 20,553 people per square mile and San Francisco has 17,180. Nassau is much denser than most places in the US at 4,704. Wikipedia puts the density of Oakland – the most urban place in the Bay Area after San Francisco – at 5,009.

            From WIkipedia, on the Village of Hempstead ( not the town ): Hempstead is just as urban (at least with regard to population density and activity) as any major city. In stark contrast to the surrounding villages in the town and county, it is more densely populated than many American cities with exception to New York City, Boston and Cambridge, Massachusetts, San Francisco, California, and Jersey City and Paterson, New Jersey.

            The City of Long Beach is denser than Hempstead Village and gets a bit closer to San Francisco densities at 16,594…

            …. Nassau is very dense, it’s just that Lawn Guylanders compare it to Manhattan and Western Queens, they think they are living in bucolic suburban splendor….

            Might want to look at


            it can be sorted by Metro Area or by State. Maybe they need to rephrase the meme and try to make parts of Nassau as dense as Great Neck Plaza instead of phrasing at as dense as Queens.

        • Nathanael

          “I’m guessing the main reasons it isn’t discussed are political and institutional. From what I understand, the current Boston political establishment is fairly anti-transit”
          …which is INSANE. What the hell is going on here?
          “—I get the impression that the MBTA gets more pro-transit pressure from Brookline, Cambridge and Somerville than Boston itself. Based on what I’ve read and a talk I attended given by a former MBTA higher-up, the T has come to see its main role in providing urban services as managing decline, not expanding the existing ridership base. ”
          Which is insane, given that the ridership base is exploding.

          Why is Boston filled with demented politicos who’ve never been in Park Street Station?

    • anonymouse

      The Yellow Line still runs to Skokie, still at grade and with crossings, but without the switch to overhead wire. There’s nothing categorically precluding grade crossings on a third rail powered line. It’s just that most such lines are also rapid transit lines, with enough traffic density to warrant full grade separation.

      • Nathanael

        Grade crossings on third rail lines require gaps in the third rail.

        This means it’s only viable for (a) narrow roads/paths or (b) long trains. No running single railcars across a six-lane road.

        And if you need long trains, you probably need grade separations.

        So the low-traffic electrifications are done with overhead. The high-speed electrifications are done with overhead. High-traffic, low-speed electrifications may be done with third rail.

        • Alon Levy

          In principle, grade crossings with third rail are possible. But they’re hard and undesirable. For example, the reason the Canarsie Line was cut to its current terminus from Canarsie Pier is that the BRT preferred to electrify lines that were connected to the subway with third rail, and there were too many grade crossings east of Rockaway Parkway, which were not worth separating. Some grade crossings remained (the last grade crossing on the subway to be removed was on the L), but having trains run at street level with grade crossings every block wasn’t worth it.

        • anonymouse

          The Orange Line runs 6 car trains of 65 foot cars, so gapping isn’t as much of a problem, and in complex junction areas with no roads involved, gapping can be an issue. Also, many third rail systems are designed with all the shoes on one car being linked, and section gaps between different supply sections being separated by a gap slightly longer than the distance between a single car’s shoes so as to avoid bridging two sections through the wiring inside a car. You really should go take a look (on Google Maps) at systems like the former Southern Railway’s DC electrification in the south of England, or Metro North and the LIRR, or even the Berlin S-Bahn. They’re all good examples of third rail systems with grade crossings, generally at the outer extremities.

  10. Pingback: Washington Union Station | Pedestrian Observations

Leave a Reply to Adirondacker12800 Cancel reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.