Skip to content

Scranton to New York in 2:59. Allentown in 2:23. Six rail routes, timed on rebuilt track.

Six routes in the Northeast, most with no passenger train today, run in a model as if their track were rebuilt and their stations reopened. Going around Philadelphia saves a Harrisburg–New York express 25 minutes. Banking curves costs a fourteenth as much per minute saved as faster track.

By Liam Blank · Published

Key findings

  1. On rebuilt track, Scranton is 2:59 from New York, Allentown is 2:23, and Bethlehem is 1:24 from Philadelphia. None of the three trips has a train today.
  2. A Harrisburg express that goes around Philadelphia reaches New York in 2:08, 25 minutes sooner than one that goes through it. The fastest train today takes 3:05.
  3. Harrisburg to Boston takes 6:15, against 7:28 at today’s speed limits.
  4. Raising 80 mph track to 110 saves more time than banking curves, 7.5 minutes against 3.0. Banking costs a fourteenth as much for each minute.

No passenger train runs between Scranton and New York. None runs between Allentown and New York, or between Bethlehem and Philadelphia. A train from Harrisburg to New York goes into Philadelphia and comes back out. The routes for something better mostly exist: freight lines, commuter lines, and a few stretches where the rails are gone and the roadbed remains.

In 2015 and 2016 I worked on a study that asked how fast passenger trains could run on six of these routes if the track were rebuilt and the old stations reopened.1 This page answers the question with today’s data: 522 curves, the railroads’ own speed limits, and trains now in service.

A note to skeptics

These are running times, not timetables. A timetable adds spare minutes so that a late train can recover. Agencies have published times for five of these stretches; run with an agency’s own stops and speed limits, this model’s time is shorter on all five: by 10 to 14 percent on four,2345 and by 25 percent on the fifth, a 2008 study that prints no speed limits.6

The times also assume the work is done: track rebuilt, stations reopened, room for the trains on the commuter lines into New York, and overhead wire for the electric train from Harrisburg to Boston. None of that is priced here. One choice inside any rebuild is priced, faster track against banked curves, because the model can measure what each buys.

Understanding the data

The routes, the stations, and most of the curves come from the 2015–16 study.1 The curves between Scranton and the Delaware River come from the survey Amtrak, the national passenger railroad, commissioned in 2023.7 Today’s speed limits come from the timetables railroads issue to their crews.89101112

Exhibit A The six routes, their stations, and every curve

Passenger trains run today Freight only No track in service NJ Transit line on to New York
Curve good for under 60 mph when banked 60 to 79 80 to 109 110 or more Station

Select a curve for its sharpness and its top speed. Select a station for the running time from the start of the route.

Three things set a running time: the speed the track is built for, the curves, and the stops. Federal rules allow passenger trains 80 mph on well-kept ordinary track and 110 on track held to a higher standard.13 Curves are a separate limit. Railroads bank them, raising the outer rail above the inner one, and the banking decides how fast a train may take the curve.14

Exhibit B Top speed through a curve, mph

Sharpness, in degrees turned per 100 feetRadiusFlat trackBanked 3 inchesBanked 6 inches
1°5,730 ft6593113
2°2,865 ft466580
3°1,910 ft385365
4°1,432 ft334657
6°955 ft273846
8°716 ft233340
Each figure includes the small margin above the balanced speed that federal rules allow any train.14 A 1° curve is a gentle bend with a radius of more than a mile; an 8° curve is the kind found in city terminals.

The model drives one train along each route. It accelerates as hard as the train can, slows for every limit and curve, and stops for a minute at each station, the time federal planners allow an intercity train at most stations.15 Where trains run today, it lands within 10 percent of the tightest published schedule on 7 of 9 stretches between Harrisburg and Boston.16 On the eighth, Springfield to Worcester, one train a day runs each way; there the model lands within 2 percent of the running time the state’s planners simulated.17 The ninth is the approach to Penn Station in Manhattan, where the model’s train has no others in its way.

Each route is run rebuilt for 110 mph with every curve banked 6 inches, the most any track owner here allows.18

How long from Scranton to New York?

2:59. The 87 miles from Scranton to Port Morris, a junction in northwestern New Jersey, take 1:29 with 9 stops on the way. 28 of those miles have no track in service.19 From the junction the train uses the line of NJ Transit, New Jersey’s commuter railroad. No train runs it end to end today. NJ Transit’s fastest trains cover it in two legs, 12 minutes to Dover and 78 from Dover to Penn Station, and the model’s train keeps that pace.16

Amtrak’s 2023 plan for the same trip says 2:50 to 2:53. It makes three stops west of the junction instead of 9, and it allows 62 to 66 minutes from Dover to New York, where NJ Transit’s fastest train takes 78.2

Exhibit C Trips with no train today, hours and minutes

Scranton to New York
2:59
Amtrak, 2023: 2:50 to 2:53
Allentown to New York by way of Hackettstown
2:34
PennDOT, 2024: 2:02 to 2:30
Allentown to New York by way of High Bridge
2:23
PennDOT, 2024: 1:49 to 2:20
Bethlehem to Philadelphia
1:24
PennDOT, 2024: 1:44 to 1:46, from Allentown

Rebuilt track, 110 mph NJ Transit’s line into New York, at the pace of its fastest trains today

Under each bar: the range an agency study gives for the same trip.

Which way from Allentown?

Two routes reach New York. North through Hackettstown, joining NJ Transit at the same junction, takes 2:34. South through High Bridge, onto NJ Transit’s Raritan Valley Line at Bridgewater, takes 2:23. The Pennsylvania Department of Transportation’s 2024 study gives two service plans for each: 1:49 and 2:20 for the southern route, 2:02 and 2:30 for the northern.20

Bethlehem to Philadelphia takes 1:24, on the line that SEPTA, Philadelphia’s commuter railroad, still runs as far as Lansdale. 12 miles of it north of Quakertown are now a trail.19 The state’s study estimates 1:44 to 1:46 from Allentown, five miles farther.20 South of Lansdale the model’s train makes 3 stops and takes 43 minutes; SEPTA’s fastest train today makes 9 and takes 56.16

What does skipping Philadelphia save?

25 minutes. A freight line called the Trenton Cut-Off leaves the Harrisburg main line near Exton, west of Philadelphia, and rejoins the route to New York at West Trenton, north of it. It is 10.7 miles shorter than the way through 30th Street, the city’s main station. Rebuilt for 110 mph, it carries an express from Harrisburg to New York in 2:08. The saving depends on that rebuild. At the speeds its freight trains are allowed today, the bypass is the slower way.

Exhibit D Harrisburg to New York, three ways

Around Philadelphia, rebuilt track 2:08
to Extonbypassto New York
Through Philadelphia, rebuilt track 2:33
to Extonthrough Philadelphiato New York
Fastest train today 3:05
through Philadelphia, on the Northeast Corridor
Both rebuilt-track trains stop at Lancaster, Thorndale, Exton, West Trenton, Bound Brook, and Newark. The one through Philadelphia also stops at 30th Street.

The fastest train today goes through 30th Street and up the Northeast Corridor, the main line between Washington and Boston, in 3:05.16

Harrisburg to Boston

The longest route runs 430 miles through Philadelphia, New York, and Springfield, without the bypass. No through train runs it today. At today’s speed limits the model’s train needs 7:28; on rebuilt track, 6:15.

Exhibit E Harrisburg to Boston, hours and minutes

Today’s speed limits
7:28
All open track raised to 80 mph
6:59
Rebuilt for 80 mph, every curve banked
6:42
Rebuilt for 110 mph, every curve banked
6:15

Of the 73 minutes saved, 29 come from raising slow track to 80 mph, 17 from banking every curve and rebuilding the freight-only stretches, and 27 from the step to 110.

Faster track or banked curves?

Those gains assume every curve can be banked 6 inches. Today it cannot. The owner of the track decides, and CSX, a freight railroad that owns 94 of the miles, sets the banking for its freight trains and does not raise it for passenger trains.21 It also holds passenger trains to 90 mph.

Under the limits owners set today, and starting from 80 mph track, raising the limit to 110 saves 7.5 minutes between Harrisburg and Boston. Banking the curves that hold trains back saves 3.0.

Exhibit F Minutes saved between Harrisburg and Boston, starting from 80 mph track

All three owners
7.5
3.0
Amtrak
3.5
1.2
CSX
3.3
0.0
MBTA
0.6
1.8

110 mph track, curves untouched (90 on CSX) 80 mph track, with the curves that hold trains back banked to the owner’s limit

Faster track wins on time. Banking wins on price: $389,000 for each minute saved, against $5.60 million.22 A second published price list, from Ohio, gives the same ratio between the two kinds of work.23

Exhibit G Cost of each minute saved, per track

110 mph track
$5.60 million
Banked curves
$389,000
OptionUnit cost per track-mileTrack-milesCost per trackMinutes saved
110 mph track$251,000166.5$41.79 million7.46
Banked curves$66,00017.8$1.17 million3.02
Prices are for one track; a two-track line doubles both. Faster track covers the open track on Amtrak, CSX, and MBTA lines now limited below 110 mph. Neither price includes the signal work that track above 79 mph also needs.

Does it hold up?

The price result holds in all 24 tests: three trains under eight sets of assumptions. Faster track costs 11 to 29 times as much per minute in every one. The time result holds in 19.

Exhibit H Minutes by which faster track beats banking, under eight sets of assumptions

AssumptionLocomotive, five coachesElectric railcars, 100 mphLocomotive, eight coaches
Today’s owner limits (the case in Exhibits F and G)+4.4+2.3+2.1
Adding the track whose owners publish no limit+9.7+5.9+5.8
No owner limits at all+10.5+4.8+5.7
CSX agrees to bank its curves+2.6+0.4+0.01
Every curve turns out to hold trains back+0.3-1.9-2.1
Both of the last two-4.0-6.2-6.6
No unrecorded curve holds trains back+9.2+5.2+5.5
Leaving out Springfield to Worcester+3.3+1.3+1.3

Shaded: banking saves more time. Banking is cheaper per minute saved in every cell.

The time result turns on 139 curves whose banking no public record shows. If all of them hold trains back, banking saves as much time as faster track, or more. The railroads’ own track charts would settle it.

The answer

The trains these routes were drawn for are fast enough to matter: Scranton under three hours from New York, Allentown under two and a half, Harrisburg 57 minutes closer than it is today. Whichever is built, bank the curves before buying faster track. Each minute costs a fourteenth as much.

Sources

  1. ReThinkNYC regional rail study, 2015–16: route maps, station lists, curve records, and trip-time spreadsheets, filed as the Green Line (Harrisburg to Boston) and the Orange Line (the shorter routes). Author’s files. ↩
  2. Amtrak, Analysis of Options for Scranton–New York Amtrak Passenger Rail Service, March 22, 2023, Figure 3. ↩
  3. North Jersey Transportation Planning Authority and NJ Transit, Central New Jersey/Raritan Valley Transit Study: Feasibility Analysis Report, April 2011, p. 30. ↩
  4. Bucks and Montgomery counties, Lansdale–Quakertown Corridor Alternatives Analysis, draft report, June 2011, Tables 6-1 and 7-3. ↩
  5. NJ Transit, Proposed Restoration of Passenger Rail Service on the West Trenton Line: Draft Environmental Assessment, Chapter 2, November 2007, p. 2-26 (archived copy). ↩
  6. Federal Transit Administration and NJ Transit, Lackawanna Cut-Off Passenger Rail Service Restoration Project Environmental Assessment, June 2008, Table 2.2-1 (archived copy). ↩
  7. Jacobs Engineering for Amtrak, Scranton–New York City Intercity Passenger Rail Analysis: Infrastructure Assessment, February 2023, Appendixes 2A and 2D: every curve between the Delaware River and Scranton, with its measured banking. ↩
  8. Amtrak, Northeast Corridor Employee Timetable No. 1, Special Instructions, effective May 4, 2026, with Summary Bulletin NYW1-31SUM, August 3, 2026: Harrisburg Line, Newark to New York, Hell Gate Line, Springfield Line. ↩
  9. Metro-North Railroad, Employee Timetable No. 3, May 15, 2016, New Haven Line speed table (NTSB docket copy). ↩
  10. SEPTA, Regional Rail Timetable No. 4, Special Instructions, effective April 6, 2008, pages revised November 1, 2009 (FRA docket copy). ↩
  11. CSX Transportation, Baltimore Division Timetable No. 4, January 1, 2005 (Trenton Subdivision), and Conrail Shared Assets, Timetable No. 10, July 1, 2013 (Lehigh Line). ↩
  12. CSX Transportation, Albany Division Timetable No. 4, November 1, 2004 (Boston Subdivision, Springfield to Back Bay). ↩
  13. 49 CFR 213.9 and 213.307: maximum passenger speed by class of track (Class 4, 80 mph; Class 5, 90; Class 6, 110). ↩
  14. 49 CFR 213.57 and 213.329: curve speed, superelevation up to 7 inches, and 3 inches of cant deficiency for any train. ↩
  15. FRA, NEC FUTURE Tier 1 Final EIS, Appendix B.05, October 2015, p. 21: a one-minute dwell for intercity trains at most stations and at least two minutes at the larger ones; 30 seconds to one minute for regional trains. ↩
  16. Published schedules, read October 2, 2026, from each operator’s timetable data feed: Amtrak, SEPTA, NJ Transit, MBTA. ↩
  17. MassDOT and VTrans, Northern New England Intercity Rail Initiative: Alternatives Analysis Report, January 2015, Appendix A, Figures 13 and 14: running time at timetable speeds with no dwell or recovery. ↩
  18. Amtrak, Limits and Specifications for Track Safety, Maintenance and Construction (MW 1000), July 1, 2023, section 57.2. ↩
  19. OpenStreetMap contributors, railway track geometry and maxspeed tags along each route, retrieved October 2, 2026 through the Overpass API. © OpenStreetMap contributors, ODbL. ↩
  20. PennDOT, Lehigh Valley Passenger Rail Feasibility Analysis, final report, March 2024. ↩
  21. FRA, Mixed Freight and Higher-Speed Passenger Trains: Framework for Superelevation Design, DOT/FRA/ORD-19/42, October 2019, p. 20 (CSX Standard Plan 2511); MassDOT, East-West Passenger Rail Study Final Report, Chapter 6, January 2021, p. 99 (separate track above 90 mph). ↩
  22. Quandel Consultants for the Midwest Regional Rail Initiative, Appendix E: Cost Estimating Methodology for High-Speed Rail on Shared Right-of-Way, April 18, 2011, sections 2.4.1 and 2.4.4 (2010 dollars). ↩
  23. Ohio Rail Development Commission, The Ohio & Lake Erie Regional Rail – Ohio Hub Study, July 2007, section 2.3.1 (2002 dollars). ↩

How this site sources and checks its work