Ruminations on Long Distance

I’m currently on an Amtrak train, headed back south from upstate New York, where I have been far away from exciting happenings in Greater Philadelphia. But I thought I’d take some time, while I’m in an appropriate spot, to think aloud about Amtrak, specifically about long-distance trains.

2013, to borrow phraseology from trite year-in-review articles, was the year in which I turned 30. The significance of which, as far as I can tell, is that I’m reaching the phase of life where overnight trips in coach are no longer worth the monetary savings. There are a few I might still do, if I was in a pinch and by myself, but this ship has, for the most part, sailed. It’s sleeper or bust.

I have a developing theory of sleeper trains, which is that they are essentially a point-to-point service. A sleeper passenger who is willing to pay a fare that is going to pay for most, or all, of her costs, wants a train that is leaving in the evening and arriving in the morning. Perhaps a short ride in daylight can cover more another market or two with the same departure, but the basic form is evening-morning. It requires two trainsets to operate the entire service. No sleeper service in the entire Amtrak system looks like this. The Auto Train, which is sui generis, comes closest, and doesn’t quite manage the late evening departure. This is largely not Amtrak’s fault, because the endpoints of trains in the National System are just so obnoxiously far apart. But the point stands that any stops made between 11:00pm and 6:00am are basically economically meaningless, and should be covered by another train that runs in daylight. If you don’t believe me, compare the passenger numbers for any station in Ohio, where Chicago-East Coast trains all pass through in the witching hour, with those from comparably-sized cities elsewhere. In the Buckeye State, trains might as well not exist.

There are plenty of trains that come close to the two-trainset overnight ideal on some part or another of their trip. For instance, the Crescent is almost perfect on the Washington-Atlanta segment, although it’s also the daylight train Atlanta-New Orleans, and lacks a daylight equivalent Lynchburg, VA-Atlanta. But all of the eastern routes have the critical problem of having at least one long segment in daylight, when sleepers are either empty or sold at uneconomic rates. Much as Atlantic City can’t pay for 12 months of services and infrastructure with 4 months of tourist revenue, sleepers can’t pay for 16-20 hours/trip of expenses with 8-9 hours/trip of revenue.

The morning arrival is the main appeal of the overnight trip; it doesn’t quite matter how fast or how slow one goes in the dark, as long as arrival is timed for when the sun comes up, before any possible same-day trip could have gotten in, and for less money than an evening flight and a hotel room at the destination. Oddly enough, despite it being by far Amtrak’s most popular market, a city that runs on its rails more than any other, and an exorbitantly expensive hotel market, there are no long-distance arrivals in New York until after 11:00am. While this makes some sense for trains coming from the south, since the tunnels under the Hudson River are at capacity during the morning rush, it discounts the possibility of a predawn arrival and a short wait in the station before reveille. It also does not explain why there is not an early morning arrival from Buffalo or Montreal via the Hudson Line and Empire Connection, which is busy but is not quite as completely crunched as the North River Tunnels.

Proper timing is about more than passenger convenience; it’s about cost containment. I touched on the expense involved in hauling around empty sleepers and extra trainsets during the day. There is also the matter of meal service. Amtrak includes diner meals in its sleeper fares, in order to ensure that its top-dollar passengers aren’t hungry and unhappy. Unfortunately, this practice leads to nearly the entirety of Amtrak’s infamous $72 million/year Food and Beverage losses. The Auto Train extends this expensive courtesy to coach passengers, in recognition of their end-to-end captive status; its financials are significantly worse for it. Conversely, a train that doesn’t leave until well after dinnertime doesn’t have to worry about serving dinner, or losing any money on it. And if breakfast becomes simple and Continental (or sandwichy), the expensive diner car and its expensive staffing requirements can be dispensed with entirely, in favor of the much cheaper and more profitable cafe car. Even switching to a voucher for a free, full Anglo-American-style breakfast stationside on arrival would be a significant savings for Amtrak’s ultrapremium commissary costs. Sleeper revenues are held down by the sheer lack of density possible per car, unless we go to Couchette or Slumbercoach setups, and long-distance trains as a whole are further hurt by the sheer number of non-revenue or low-revenue cars they must carry in addition to sleepers and coaches. It’s telling that, of the 130 cars in the initial Viewliner II order, only 25 are pure sleepers. It’s not that the baggage cars and diners are unimportant to Amtrak; they’re critical. (Most of Amtrak’s current bag cars and single-level diners predate its own 1971 founding.) But they will help Amtrak’s bottom line by saving expenses, not generating revenue.

So what I’d really love to see from Amtrak are sleeper trains that provide an extra option to the existing middle-distance corridors, especially the ones that have one end in New York. Buffalo-New York, Montreal-New York (agnostic as to Plattsburgh vs. St. Albans), Pittsburgh-New York, Raleigh-New York, Norfolk-Philadelphia, and Cleveland-Philadelphia, are all about the correct distance for a point-to-point sleeper service. Boston-Washington had sleeper service very recently, and Amtrak has all but explicitly promised to restore the solitary sleepers to trains 66 and 67, which would also restore sleepers-without-diners to the Amtrak system, in case anyone at 60 Mass Ave needs the reminder that it can be done.

Not all of those are equally ripe for startup today, and each comes with unique challenges. Buffalo-New York is almost ideal, with four daylight round trips already on the Empire West schedule, but nobody is going to resist the temptation to extend it to Chicago, giving Ohio its first daylight service, and given the insanities of the current Lake Shore Limited service, it’s not clear they should. That also avoids creating a new equipment base, which saves rolling stock just as much as aggressive short-turning. Similar for Philadelphia-Cleveland, although that route has zero direct trains today. The services coming from the NEC South all have to be navigated around the aforementioned blocked-out morning rush, either by sneaking in to Penn Station before the rush, or being held at 30th Street until after. Montreal only has the Adirondack and needs to finish its border preclearance station, although a sleeper schedule is a good way to get further use out of that investment.

Unfortunately, the regulatory environment for new sleeper services is complete hash, even as Amtrak seems to be entering an era of steady-state production of long-distance cars at the CAF assembly plant in Elmira. Some routes, like New York-Buffalo-Chicago, are longer than the 750-mile legal definition of long distance set out in PRIIA ’08, and are therefore at Amtrak’s discretion. But many of the routes I listed are shorter than that, and are required to have one or more state sponsors, which is going to be a very difficult political lift for what is basically a premium service. At least a state might be more willing than Amtrak itself to experiment with dinerless sleepers, especially New York on the Montreal route, whose anti-F&B frugality led to the death of cafe cars on the Empire South trains.

It will be a few years before Amtrak has to worry about any new sleeper service; it won’t have more than a handful of extra sleepers until 2015, and they are already earmarked to 66/67 and to lengthening existing trains, which is a very cost-efficient expansion of capacity. Sleepers also need to be accompanied by coaches, for the bottom (i.e. under-30) segment of the market, and Amtrak is short of single-level coaches for the foreseeable future. But if intercity rail has a future in America, then at least some of that will be long distance rail with attached sleeper cars; that is one sense in which we really are a spread-out nation. It’s worth thinking about ways in which the next generation of sleeper service can be better and more self-sustaining than the present one.



Several of the thoughts in this post have been percolating in my brain for a while, but specific ideas about where the revenue and expense centers are on a long-distance train were provoked by Paul Druce, who blogs at Reason & Rail, and has been running an occasional series on Amtrak financial performance, which is very much worth reading, even though he is coming from a very different place and reaches different conclusions than I do. Other positions included in this post were reached in view of the performance of Western Europe’s remaining sleeper services, especially DB’s City Night Line and ScotRail’s Caledonian Sleeper. It’s been fifteen years since my one ride on a CNL train, so I relied heavily on British rail expert Mark Smith, a/k/a The Man In Seat Sixty-One for guidance as to what’s worked and what hasn’t, in an environment where governments have been investing heavily into high speed rail and discount airlines have attacked the point-to-point long-distance market with a vengeance.

Join the Conversation


  1. Very good analysis. Rather than Buffalo-NYP, though, wouldn’t Toronto-NYP be a more likely successful sleeper route? It’s only a couple of hours further (assuming border issues can be worked out), and is a much bigger market with no great non-flying travel options today. The strength of Empire Service west of Albany isn’t the Buffalo-Niagara area as a terminal itself; it’s the concentration of midsize cities (Utica, Syracuse, Rochester) along the route. A sleeper train would hit those at Ohio-esque times, and is Buffalo really enough of a ridership anchor on its own? My guess is that a train leaving Toronto early evening, stopping in Buffalo before 10, and then on to NYP by 9 AM would do very well. Of course, the real game-changer for the west-of-Albany corridor would be for someone to find the money to third-track the whole route….

    1. Well, first of all, I’m phobic of border crossings because they destroy OTP unless you can put the bidirectional port of entry at the terminal, like they’ve done at Vancouver and is on the table for Montreal. Toronto-Niagara Falls is never going to be amenable to that solution.

      My thought process on Empire West are as follows. You’re right that the string of similarly sized cities is what makes this such an attractive corridor for daylight service. But it’s also true that the quality of service degrades as you proceed west, just because the Big Travel Market is NYC, and Buffalo is just farther than Rochester which is farther than Syracuse which is farther than Utica. Even this trip I’m on right now from Rochester is basically eating my entire day. And the requirement that both endpoints of a 9 hour trip (NFL-NYP) have to be during waking hours, even loosely defined, mean that the possible trip times are weirdly bunched up in a relatively small slice of the day; on weekdays, the last eastbound train from Rochester leaves at 2:13pm. Not the most flexible corridor schedule. A sleeper BUF-NYP at least gives an alternative to that. And if you doubt the ability of Buffalo/Niagara (with an assist from Rochester) to fill a train, then Cleveland and Erie are just as good as Toronto to help put souls in bunks, without playing CBP Roulette. And either way, if you terminate at either Buffalo or Cleveland, you need to park a third trainset there as equipment protection, so then you might as well use that extra trainset to run all the way to Chicago, and reuse the protection you already have there for the Lake Shore. (Going beyond Buffalo in either direction costs you the option to go dinerless, but the political oomph from hitting Cleveland at a sane hour, combined with the legal inability of Amtrak to launch CHI-CLE service on its own accord, is probably worth it.) And even Chicago might appreciate the ability to choose to arrive on the East Coast in the morning and not midday or dinnertime.

      Relieving the many track bottlenecks on the Empire West would indeed be a blessing unlooked for. As of now, though, my cynicism regarding CSX’s friendliness is nearly unbounded.

      1. If you wanted to run to Chicago, you could still do so without the diner by switching out the sleepers at Buffalo. It would cost ~15-30 minutes and a few million for a storage track, but you’d save money on not having a diner, its crew, potentially the crew of the sleeper cars (really not much of anything for them to do BUF-NYP), and not carrying the cars to/from Chicago (~400K per year per car). Downside is that you lose sleeper traffic Cleveland-New York, but I suspect that the gains are worth it. On the other hand, any Buffalo-only sleeper is going to be leaving late in the evening unless you want to arrive absurdly early, whereas Cleveland has a fairly decent hour of arrival/departure. This brings to mind another option, a “coach-only” train which uses lay flat or open suites for business class (which would probably need to bring back the old Custom Class designation). This still avoids diners and all their financial misery, while still allowing for happily up charging and probably with more capacity as well (the new Viewliners can only hold 22 passengers iirc).

        1. Sleepers leaving late in the evening (or arriving early in the morning) shouldn’t be a problem — the set-out sleeper solved that problem a century ago. (Amtrak was even using them on 66/67 at NYP for a while, weren’t they?)

          If the sleepers are on the rear end of the consist and the storage track had spurs at both ends, no switch engine or engine crew would be needed for the switching. Heck, this way you could keep some sleepers through to Cleveland while dropping others at Buffalo. Not too different from the MSP or DEN sleepers, except that with the correct sort of storage track you don’t need an engine crew at BUF.

          I have a hard time imagining sleeping cars without car attendants, though, so there would need to be some crew accommodations at BUF, as well as a car cleaning crew. Not a huge deal, though.

        2. 1. I’m really wary of killing diners entirely on any route of longer than eight hours or so, especially when they run during obvious mealtimes, which a New York-night-Bufalo-day-Chicago train would. And while the westbound departure from NYP and eastbound departure from Buffalo would be timed to be well after dinner, the eastbound passengers from Cleveland are going to be rather hangry with just the cafe available until dawn.

          2. The reason to run through to Chicago in the first place would be to reduce or eliminate the need for an additional fixed base of employees or equipment at Buffalo. Setout sleepers and other such kludges defeat that purpose. Maybe if the level of traffic is sufficiently high at a later date, but with Western New York’s terminal forces split between Buffalo-Depew and Niagara Falls to begin with, I think this doesn’t work. It might have a better shot at Pittsburgh than Buffalo, but there are reasons why Amtrak doesn’t do midline switching anymore (except at Albany, Spokane, and San Antonio).

          3. Not precisely sure avout the Viewliner IIs, but on Amtrak’s Viewliner I floor plan, I count 3 bedrooms and 12 roomettes for a theoretical maximum of 33 passengers, but that requires every compartment at full occupancy, which Amtrak even recommends against. That’s still awfully low, and you can see from the other commenters that we’re not the only ones facepalming over it. I’ve casually considered a Slumbercoach-type sleeper based on NJT/AMT’s MultiLevel coach, but I haven’t sat down with a spec sheet to do math yet. It would possibly be more pain than it’s worth, since you couldn’t have it in the middle of the consist without falling afoul of ADA.

          1. There are also the Denver and MSP sleepers, though, which is more akin to what I was suggesting with Buffalo-sleepers on a NYC-Chicago run. Baggage, 3 coaches, cafe, diner, 2 NYP-CHI sleepers, 2 NYP-BUF sleepers, something like that. Still need at least a minimal BUF crew base for car attendants and car cleaners, though. So what’s the cost/benefit of that versus just running those two sleepers through to CHI?

            What’s the ADA foul?

            1. Are there still DEN and MSP sleepers? I was under the impression that those had gone the way of the dodo.

              Establishing a crew and equipment base at Buffalo is a big deal; we’re talking at least a few million dollars up front for yard track, car barn, switcher (I’m going to assume one will be needed until proven otherwise), plus ongoing expenses like new headcount. My favorite contact inside Amtrak is phobic of new unfunded bases more than anything else; he doesn’t publish numbers publicly, but seems to have his head generally screwed on right, so I’m kind of trusting him that it’s really the order of magnitude of ongoing expense he implies.

              I know Amtrak has been contemplating a sleeper, a coach, or both for ATL on the Crescent, but hasn’t yet despite even more lopsided ridership than on the EB east of MSP or the CZ east of DEN. (The recent EB surge in Williston-MSP traffic has smoothed that out some…) Similarly, running through-cars between the Pennylvanian and Capitol Limited at Pittsburgh has been thoroughly considered; again, still no movement yet. The expense of setting up switching crews has been pointed to as the dealbreaker, at least for now. Maybe Amtrak is too conservative on the subject, but at the very least I don’t want to second-guess them without much better data than what I have.

              That said, if BUF sleepers were to be a realistic thing, I like your proposed consist, with the possible amendment of more coaches up front. But I should also bring up the elephant in the room: Boston. A second LD on the Water Level Route might call for a Boston section much like the first, and that would involve a lot of mucking around with consist management if there were also BUF sleepers. But a Boston section might not fit the model of service we want either; If the westbound New York section leaves at a sufficiently late time to avoid dinner in the diner, but 449 leaves Boston 3h45m before 49 leaves New York. You’re stuck serving dinner on a Boston section if you have one. It might be more worth it to make BOS passengers connect on 177 (wb) and 2150 (eb). SPG would need Shuttles added to the schedule, and WOR would just be out of luck, but can’t have everything, can you?

              The other option for BOS is to go fully CityNightLine, and have a lot of consist surgery happen overnight at Rensellaer: nb from New York meets a wb from Boston, they swap cars, and continue to both Chicago and Montreal. Shortly thereafter, the eb from Chicago and the sb from Montreal meet and reverse the process. The diner goes from Chicago to Boston and back. This plan involves timekeeping reliability that Amtrak has never gotten from CSX either east or west of Albany, never mind the D&H, so it’s impossible, but if American freight railroads ever discover German Punctuality (I once received an apology for a 90 second takeoff delay on Lufthansa), it would be an interesting proposition.

              The ADA foul is not being able to get through the ML car to other critical points on the train, like the cafe/lounge or the diner. Not a problem on commuter rail, where the train is uniform and the only level piece of floor needs to be between the door and a seat, but intercity is different.

            2. Note on the ADA issue: there are acceptable workarounds that would make the cars kosher in the eyes of the law (the Superliners are legal, after all), but it’s a huge can of worms that I would need to feel a lot better about before opening.

  2. Nice article, definitely agree. I really want to take a train from Philly to Cleveland, but there really are no good ways to do it without having to be somewhere at 3am.

  3. Oh and definitely would love to see a sleeper from New York to Montreal. At this point it’s drive or nothing to get to Canada, because flights are so comically expensive. I’d love to take a sleeper up there.

  4. You’re on 64? I’m on 49 heading north, so I’ll wave to you in Albany. 🙂

    I’m a bit dubious about the profitability of an overnight Adirondack or Vermonter. While most of the intermediate stops don’t have very many boardings, they do add up when taken together, and you lose all of those (except maybe as far as PLB/ESX southbound, or POU/NHV northbound ) if they’re in the middle of the night, especially the SDY-NYP segment of the Adirondack, which is basically another Empire Service train at that point.

    1. I was on 284, so you missed us around Poughkeepsie.

      You have a fair point about losing the intermediate stops. My counterargument would be that endpoint-endpoint is still the most popular single city pair on the route, despite the truly egregious schedule. Most of the rest of the top ten city pairs include Montreal. That should be evidence that Montreal is a popular origin/destination, and there are NYP-MTL passengers who are flying or taking other options because the schedule is too unfriendly. Also, a night schedule allows for connections, both PHL/WAS-MTL and NYP-Ottawa/Quebec, which the current schedule… is much less friendly towards. An overnight schedule doesn’t help Saratoga-Montreal passengers, but the point is that they already have the Adirondack.

    1. I liked the dinners on the Southwest Chief! I don’t want to get rid of those, they’re good!

      And I paid actual cash money for those dinners, because I was young and cheap and in coach, but despite the high prices I paid, Amtrak still took a big loss on that meal. *sigh* I don’t know how to fix the western transcontinental trains, I really don’t. The distances involved are too long for the kind of tinkering I was contemplating here. At least city-pair distances are saner on this side of the Mississippi.

      1. My dining car experiences (multiple EB, CZ, and SWC round-trips, plus a gazillion Crescents and some Silvers) are that the crew makes a huge difference in whether Amtrak maximizes its revenue potential from a diner or not. (Note, I’m aware that even maximizing potential does not necessarily equal breaking even or making profit. But it’s better than the alternative.) Good crews motivate passengers in coach to use the dining cars, efficiently seat and serve multiple sittings per table for every table for every meal, and generate return business through customer satisfaction and word-of-mouth. Bad crews barely make any effort to tell coach passengers that there’s a food option other than the cafe, occupy multiple tables for crew-chit-chat and busing stations, don’t optimize seating at tables, and don’t take multiple reservations per table per meal. A good dining car crew sells easily three times as much food per meal, in my experience … and creates a better ride experience for more passengers, building positive branding. Amtrak should at least start by addressing this on their long-distance routes.

          1. But are they in the position of increasing their overall loss on every sale? I don’t think so — I think greater volume helps them reduce their loss.

            I’m under the impression that any additional meal itself — just the food itself, preparing it in the commissary, storing it on the train, prepping on the train — generates money if you look at those costs against menu prices. Whether you’re serving 40 dinners or 80 dinners or 120 dinners in a Viewliner Diner, you’ve got fixed costs for the diner, maintenance, and crew, and additional linens aren’t really an added cost worth thinking about. I think the diners have the storage capacity, the single chef has minimal prep-work per meal given what the commissary provides these days and there’s no onboard dishwashing. (Sure, the second person in the kitchen would help, but I’ve seen good crews manage high-volume diners without.) So more meals per trip means less money lost per trip, no?

  5. I’ll be riding CNL Copenhagen-Amsterdam (and return) next month. I don’t understand why we can’t have couchettes here. I mean, I’ve heard the argument is “Americans can’t share space with strangers”, but, please, really? These point-to-point overnight trains should be baggage, coach, three couchette compartment cars, cafe/lounge, three sleepers as consist.

    1. The Viewliner I carries a maximum of 33. The Budd Slumbercoach carried 40. The Siemens DB couchette carries a maximum of 60. Honestly, I’d still prefer the Slumbercoach, but unless there’s a future Slumbercoach II that carries 48 or more, I will acknowledge that the couchette wins.

Leave a comment

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

%d bloggers like this: