Pennsylvania 4-4-4-4, 6-4-4-6, etc. "Duplex Drive" Locomotives in the USA


ClassQty.Road NumbersYear BuiltBuilderNotes
  1. Numbers 6110 and 6111 dropped from roster in 1953 and scrapped
  2. Number 5508 dropped from the roster in 1952 and scrapped. The balance were dropped in 1953 and scrapped.
  3. Numbers 5525-5529 dropped from the roster in 1953 and scrapped.
  4. Numbers 5517 and 5522 dropped from the roster in 1952. The balance were dropped in 1952 and scraped.
  5. Numbers 5530-5541 dropped from the roster in 1953. Numbers 5542-5549 were sold for scrap in 1956.

Class Details by Steve Llanso of Sweat House Media

Class Q1 (Locobase 349)

Data from Pennsylvania data card and photo E.13377 reproduced on []. See an extended, entertaining dialogue among several thoughtful ferro-equinologists on duplex setups in general and the Pennsy engines in particular at's thread [] (last accessed on 22 March 2015). Begun by Allen Hazen on 7 August 2007, the discussion ranged over several different US duplex designs and delved in great detail. (Thanks to Chris Hohl for the valve gear ID.)

Only one Q1 was built in 1942. The duplex-drive idea, having smitten the Pennsy in the passenger field, took hold of its freight engine designers, too. The Q1's drivers were tall for a freight engine. Its rear two cylinders were reversed under the cab (rods moving forward) to drive the two axles of the second group on the rigid frame. The cylinders sizes and strokes were unalike, which except for compounds, is so unusual as to be unique. The front pair had 23-in diam pistons with a stroke of 28 inches, the rear pair had 19 1/2-in diam pistons with a shorter 26-in stroke.

After further testing, the Pennsy decided to place the two-coupled axle forward and the three-coupled axle in the rear, and to duplex-drive them with conventionally arranged pistons. This resulted in the Q2, described in Locobase 350.

Class Q2 (Locobase 350)

Data from PRR Steam Locomotive Diagrams supplied in May 2005 by Allen Stanley from his extensive Rail Data Exchange. Additional comments from Don Ball, Pennsylvania Railroad 1940s-1950s (Chester, Vermont: Elm Tree Press, 1986), pp. 28-29. See an extended, entertaining dialogue among several thoughtful ferro-equinologists on duplex setups in general and the Pennsy engines in particular at on duplex setups at's thread [] (last accessed on 22 March 2015). Begun by Allen Hazen on 7 August 2007, the discussion ranged over several different US duplex designs and delved in great detail. Data for tender details from Classification and Description of Locomotives and Tenders, No 109-J, Pennsylvania Railroad, May 15, 1948, p. 46. (Thanks to Chris Hohl for his 11 August 2017 email noting the tender weight.)

Pennsy works numbers were 4527 in August 1944, 4535-4536 in January 1945, 4537-4539 in February, 44540-4543 in March, 4544-4548 in April, 4549-4554 in May, and 4555-4559 in June.

It's difficult to see what the Pennsy gained in designing this duplex-drive locomotive so late in the steam locomotive's day. (See Locobase 349 for the single Q1 trial horse.) It is possible that the lesser weight of each driving set (cylinder, main rod, side rods) allowed smoother running at higher speeds. It is, in most respects, virtually identical in size, power, and weight to the J-class 2-10-4s that Pennsy built according to an earlier C&O design, although the tractive effort was 5.5% higher.

The Q2 was a rigid-wheelbase locomotive that was subdivided into two axle groups: the forward pair of axles were driven by 19 3/4-in diameter cylinders with a stroke of 28 inches. The rear cylinder pair, located between the front axle and the three-coupled rear group, had 23 3/4-in diam cylinders with a stroke of 29 inches. A trailing truck booster added 15,000 lb (6,804 kg) to the already high starting tractive effort. The engines were large and attractive with an air-smoothed casing similar to that of the SP's 2-8-8-4. 26 were built, including 6131 and the 25 built in 1945-1946.

The Q2 was credited with having developed 8,000 peak horsepower (5,966 kW). In tests, the Q2 reached 7,987 hp at 57.4 mph evaporating 16,600 US gallons (62,831 litres) of water per hour and burning 12.5 tons (11.3 metric tons) of coal. According to Don Ball, the Q-2 was a powerful machine in actual service as well, but he did not think the power worth the Q2's liabilities:

"Things get interesting, however, when you realize the Q2 is, in effect paying a dead weight penalty of 43,220 pounds [19,604 kg] for 1000 more indicated horsepower(746 kW) at 57 miles per hour [than that developed by the J1]." He concedes that piston thrusts were lower but at freight-train speeds the J1 wouldn't encounter much stress.

Against those modest positives, Ball arraigned the Q2 on several charges:

"The Q2's did not know how to conserve water and in short time enginement were told to plan 'no more than an hour and a half of hard running before needing water'". They also were maintenance hogs: "When all the Q2s developed leaks in the barrel seam of the boilers just ahead and back of the second set of cylinders [probably due to frame stress], extensive work was required to do the caulking of the boilers."

Class S1 (Locobase 347)

See also "High-Capacity Locomotive for Fast Service", Railway Age, Volume 106, No 25 (24 June 1939), pp. 1067-1075, 1079. (Thanks to Chris Hohl for his 3 January 2018 email that included a link to the long S1 article noted in this paragraph, correcting the wheelbase measurement, and pointing out the separate exhausts.)

Passenger locomotive design run wild was the S1. Chris Hohl pointed Locobase to a previously unknown (to Locobase at least) collaboration between the Pennsy and American Locomotive Company, Baldwin, and Lima to design this engine. It was built in time for the 1939 New York World's Fair, where it was displayed on a treadmill operated by its own steam. COD price was $669,780. Given its destination and the combined efforts of four of the most important steam locomotive design organizations, this engine was clearly a showpiece.

Note the very tall drivers, odd-ball three-axle leading and trailing trucks, and large tender. A Commonwealth steel bed casting integrated two pairs of cylinders and its valve chambers for 12" (305 mm) piston valves with back cylinders and valve-chamber heads, front cylinder saddle supporting the smokebox. Each pair of cylinders exhausted into its own blastpipe, the rear stack emerging behind the front at the top of the smokebox.

The large Belpaire firebox at one end met the Pennsy's standards; its heating surface area included that supplied by seven American Arch circulators. Water that passed through 5 1/2" (140 mm) horizontal tubes met at the centerline with the other crosstube forming the bottom of the 7" (178 mm) vertical tube that sprayed the water up into the steam space above the crown sheet. The lowest set pair of tubes was forward with the side openings of the other six steadily rising toward the back. A large Worthington 6 SA feed water heater was fitted with a 7 SA pump to handle the enormous boiler's thirst.

The tall Baldwin disc drivers turned in Timken roller bearings. Each leading driver (axles 1 and 3) enjoyed some lateral play in boxes fitted with Alco lateral motion controls.

The usual tractive effort figure is given as 71,900 lb (32,613 kg).at 80% cut-off. The figure given in the table permits easier comparison with other engines, most of which have indicated TEs based on the 85% efficiency factor.

Unfortunately, this exceptionally long engine and tender combination, styled by Raymond Loewy, was prone to slipping and was much too large for most service. Staufer (1968) agrees that she was oversized and thus unable to visit most roundhouses or handle tight curves, but contends: "She was an excellent steamer and gave trouble-free service." Indeed, this engine is regarded as the unofficial speed record-holder for steam by many because of runs she reportedly made pulling 850 to 1000-ton trains at 140-145 mph (225-233 kph).

Note that this high speed is made pulling a weight of train most other passenger locomotives couldn't get up to 50 mph (81 kph). On the other hand, such claims have to answer how the design could handle piston speeds of 2,400 fpm (733 mpm), drivers rotating at 560 rpm (9.3 rev/sec) and a cylinder demand of 25,600 cu ft/min (426.7 ft/sec)

Class T1 (Locobase 348)

Data from tables and diagrams in 1947 Locomotive Cyclopedia, supplemented by PRR Steam Locomotive Diagrams supplied in May 2005 by Allen Stanley from his extensive colleciton. See also an extended, entertaining dialogue among several thoughtful ferro-equinologists on duplex setups in general and the Pennsy engines in particular at on duplex setups at's thread [] (last accessed on 22 March 2015). Begun by Allen Hazen on 7 August 2007, the discussion ranged over several different US duplex designs and delved in great detail. A Trains Forum thread included locoi1SA's 3 March 2012 repost of John R Crosby's "Last Chance for a Pennsylvania T1" (originally published in Trains Magazine) at [], last accessed 25 November 2018., (Thanks to Chris Hohl for querying various data points.)

Baldwin's works numbers were 62451-62452 in 1940 and 72764-72766 in November 1945; 72767-72768 in December; 72769 in January 1946; 72770-72772 in April; 72773-72776 in May; 72777-72780 in June; 72781-72784 in July; 72785-72788 in August.

The Pennsy's own Juniata shops turned out 5500-5524 simultaneously; their works numbers were 4560-4563 in November 1945, 4564-4568 in December, 4569-4570 in January 1946, 4571-4574 in February, 4575-4577 in March, 4578-4580 in April, 4581-4582 in May, and 4583-4584 in June.

The first two engines for the Pennsy in this arrangement were the 6110-6111, built by Baldwin in 1942. An February 2011 email from Mr Andrew Fields pointed out that the only difference between the two was that the 6111 had a Franklin booster engine on the rear truck that added 13,500 lb to the starting tractive effort.

Most sources including surviving locomotive diagrams prepared by the Pennsy itself give the 1,430 sq ft superheater area shown in the specs, but it's worth noting that the 1947 Locomotive Cyclopedia credited the superheater with 1,680 sq ft (156.05 sq m) on page 213.

Setting the style for the fifty that followed (many felt prematurely) in 1945-1946, the 6101s had Franklin A oscillating-cam poppet valve gear, Raymond Loewy styling with a chisel nose, disc drivers, and most unusually, a rigid wheelbase of four axles divided into two groups, each of which was driven by a pair of cylinders.

The theory behind the duplex drive held that reducing the masses of reciprocating (e.g., main and side rods) and revolving (e.g., cranks and counterweights) would reduce hammering and raise speed. Tests run at a high boiler pressure and late cut-off showed 6,110 drawbar horsepower at 85 mph. In service, these engines could be fast, powerful, and smooth, pulling 910-ton passenger trains at 100 mph.

According to most sources, the divided drive proved to be a major maintenance headache, however, because no way could be found to stop one or the other driver-cylinder set from slipping, either starting or, more alarmingly, at high speed. This liability proved fatal to the chances for real success for these engines, and they were retired well before the much earlier vintage K-4 Pacifics.

PRR T1 Locomotive Trust's FAQ--[], last accessed 9 June 2020--ascribes the wheel slippage isue to two root causes-ineffective spring equalization and improper handling. The equalization problem was solved in production engines, says T1, by splitting the equalization. Establishing the proper way to handle the T1's . Moreover, the trust FAQ offers the following steps it will take:

"We will be performing kinematic and compliance simulations of the spring rigging and equalization to determine whether further improvements in adhesion are possible. We will be applying a wheel slip alarm, so the engineer would be made aware of a wheel slip more quickly should it occur, and reduce power manually. We will also investigate fitting an electro-mechanical anti-slip device similar in concept to that fitted to the Q2, but with more reliable valves and modern electronics, so no involvement from the engineer would be required."

Another source of difficulty was the Type A oscillating-cam system that operated the poppet valves. Train Forums contributor David Stephenson, writing as "feltonhill" in a post dated 21 August 2007 ([]), described the mechanism: "Oscillating cams took their motion from a gearbox containing what amounted to a pair of miniature Walschaert valve gears. The valve drive cam moved through a small arc but did not fully rotate 360 degrees. Cutoff was controlled within the gearbox mechanism."

Pointing out that other oscillating-cam systems were installed in accessible places on locomotives, feltonhill reports that the T1's installation was: "...a mechanic's nightmare. Although the drive boxes themselves were considered fairly reliable, their location was about as inaccessible as one can imagine and routine mainenance suffered accordingly. The front box was under the "hood" of shrouding on the pilot deck, behind the aftercooler and between the air pumps ...or the rear engine set, the box was located behind the cylinder saddle and positioned vertically between the frames. Nice place to work. Oh, and just for further complication, it was a mirror image of the front drive box."

5500 was converted to a Franklin Type B rotary-cam valve system in July 1948 and 5547 was fitted with Walschaert gear in July 1949. The rotary-cam drive's cam "...operated more like those in IC engines, i.e., it rotated 360 degrees and the valve timing was controlled by the contours machined into the cam." Both conversions are reported to have been more reliable and maintainable.

Nine of the engines later had their cylinder diameters reduced by an inch to 18 3/4" (470 mm).

Stephenson offered a more complex view of the design in the May 2005 Chesapeake and Ohio Historical Magazine (reproduced online at [], accessed 24 Oct 2005):.

He says that C & O tests of the T1 in regular passenger service show that:

"They handled trains well, particularly at higher speeds.

"They kept schedule and made up delays on most runs.

"They had no excessive tendency to slip.

'The stall at Waynesboro [a September 12, 1946 event often offered as a primary example of the design's slipping tendency] was caused by overloading."

Stephenson's meticulous, and thankfully clearly written, account of the September 1946 tests nevertheless reports some design weaknesses, particularly in starting trains. A long response in the Trainorders discussion of Caprotti valve gear include PRR engineer John Roller's method for dealing with wheel slip: "[H]e could control the PRR T1's slipperiness by letting the T1 boiler pressure come way down during station stops, helped by the injector. Upon departure time, the T1 could be eased out of the station, sticking to the rails, and as speed came up, so did boiler pressure! He had no trouble with the T1s!"

Ultimately, Stephenson argues that the duplex solution was conceived to redress a potential problem that never quite materialized. As other designers of 4-8-4s came to grips with counterbalancing and stress issues, they came up with such classics as the UP FEFs (Locobase 284), the Santa Fe 2900s (Locobase 271), the N & W Js (Locobase 274), and ultimately the S-1 Niagaras of the New York Central (Locobase 5582). "Here was a locomotive," Stephenson 's verdict concludes, "that could match the T1 at all but the highest speeds, and do it day-in and day-out without special treatment."

Ultimately, however, the T1s' early departure from service came not from their own peculiar shortcomings, says Stephenson, but from the triumph of the diesel's "...superior economics ...and their immediate application to PRR's heaviest and most prestigious trains ...In the face of dieselization, the T1 just didn't matter."

But a pass-along in the forum (archived at [], last accessed 20 August 2017) relates John R Crosby's invigorating turn as fireman in Last Chance for a Pennsylvania Railroad Class T1. Sooty and dark in its neglect, the locomotive came to life as it hustled unnamed No. 43's 15-car train (all heavyweights) over 27 miles (43.5 km) of dead-straight track at an average speed of 97 mph (156 kph). Crosby claimed to have seen 108 mph (174 kph) on the speedometer, then a pegged-to-the-stops 120 mph (193 kph) along the way.

The hell-for-leather ride demonstrates how a skilled engineer (and fireman) could work with the T1's unarguable strengths and minimize its weaknesses to turn in an impressive performance.

Plans to build another T1 (5550) took tangible form in 2014 with a fundraising goal of $15 million and a target date of 2030.. One milestone was the ordering of three boiler courses from Continental Fabricators of St Louis in January and May 2018 followed by the front tubesheet in March 2019 and smokebox order in December 2019.. One of the key changes in the design from the original is converting construction from rivets to welding.

Principal Dimensions by Steve Llanso of Middle Run Media

Locobase ID349 350 347 348
RailroadPennsylvania (PRR)Pennsylvania (PRR)Pennsylvania (PRR)Pennsylvania (PRR)
Number in Class126152
Road Numbers61306175+61005500-5549
Number Built126152
Valve GearWalschaertWalschaertWalschaertFranklin poppet
Locomotive Length and Weight
Driver Wheelbase (ft / m)26.83 / 8.1818.12 / 5.5215 / 4.5713.83 / 4.22
Engine Wheelbase (ft / m)54.83 / 16.7153.46 / 16.2964.33 / 19.6151.92 / 15.83
Ratio of driving wheelbase to overall engine wheelbase 0.49 0.34 0.23 0.27
Overall Wheelbase (engine & tender) (ft / m)107.83 / 32.87107.62 / 32.80123.76 / 37.72107 / 32.61
Axle Loading (Maximum Weight per Axle) (lbs / kg)73,700 / 33,43079,780 / 36,18871,680 / 32,514
Weight on Drivers (lbs / kg)354,700 / 160,889393,000 / 178,262281,440 / 127,659279,910 / 126,965
Engine Weight (lbs / kg)593,500 / 269,207619,100 / 280,819608,170 / 275,862502,200 / 227,794
Tender Loaded Weight (lbs / kg)412,900 / 187,289422,000 / 191,416451,840 / 204,951442,500 / 200,715
Total Engine and Tender Weight (lbs / kg)1,006,400 / 456,4961,041,100 / 472,2351,060,010 / 480,813944,700 / 428,509
Tender Water Capacity (gals / ML)18,000 / 68.1819,200 / 72.7324,230 / 91.7819,200 / 72.73
Tender Fuel Capacity (oil/coal) (gals/tons / Liters/MT)37.50 / 3437.50 / 3426.50 / 2442.60 / 39
Minimum weight of rail (calculated) (lb/yd / kg/m)118 / 59131 / 65.50117 / 58.50117 / 58.50
Geometry Relating to Tractive Effort
Driver Diameter (in / mm)77 / 195669 / 175384 / 213480 / 2032
Boiler Pressure (psi / kPa)300 / 20.70300 / 20.70300 / 20.70300 / 20.70
High Pressure Cylinders (dia x stroke) (in / mm)23" x 28" / 584x71119.75" x 28" / 502x71122" x 26" / 559x660 (4)19.75" x 26" / 502x660 (4)
Low Pressure Cylinders (dia x stroke) (in / mm)19.5" x 26" / 495x66023.75" x 29" / 603x737
Tractive Effort (lbs / kg)81,794 / 37101.18100,816 / 45729.4276,403 / 34655.8664,653 / 29326.14
Factor of Adhesion (Weight on Drivers/Tractive Effort) 4.34 3.90 3.68 4.33
Heating Ability
Tubes (number - dia) (in / mm)52 - 2.25" / 5751 - 2.5" / 64219 - 2.25" / 57184 - 2.25" / 57
Flues (number - dia) (in / mm)236 - 3.5" / 89277 - 3.5" / 8969 - 5.5" / 14069 - 5.5" / 140
Flue/Tube length (ft / m)20 / 6.1021 / 6.4022 / 6.7118 / 5.49
Firebox Area (sq ft / m2)580 / 53.88725 / 67.35660 / 61.32490 / 45.52
Grate Area (sq ft / m2)121.70 / 11.31121.70 / 11.31132 / 12.2792 / 8.55
Evaporative Heating Surface (sq ft / m2)5518 / 512.836725 / 6255661 / 526.124209 / 391.03
Superheating Surface (sq ft / m2)2290 / 212.832930 / 272.302085 / 193.771430 / 132.85
Combined Heating Surface (sq ft / m2)7808 / 725.669655 / 897.307746 / 719.895639 / 523.88
Evaporative Heating Surface/Cylinder Volume409.87677.22247.42228.25
Computations Relating to Power Output (More Information)
Robert LeMassena's Power Computation36,51036,51039,60027,600
Same as above plus superheater percentage47,09847,46350,29234,500
Same as above but substitute firebox area for grate area224,460282,750251,460183,750
Power L149,78151,08250,89642,362
Power MT1547.061432.781594.751334.60

All material Copyright ©
Wes Barris