The Trainphone: PRR’s Wireless Radio System

Last revised: September 22, 2026

The bar on the roof of a Pennsylvania diesel is the Trainphone. It looks like a handrail because that is what the shop built: an insulated tube, with the transmitting wire inside it, running the length of a cab unit, a road-switcher cab, a steam tender, or a cabin car. Union Switch & Signal developed it with the Pennsylvania Railroad. The circuit was an inductive telephone. A handset in the cab put a voice on a low-frequency carrier. The magnetic field around that roof bar induced a current in the rails, or in the open wires of the pole line, and those conductors carried the talk to the other end of the train and to the block operator.

It was the railroad’s answer to the hoop and the stop. A towerman with a train order tied to a hoop had one chance as the engine passed. A red board that made the train stop for the order cost a schedule. The Trainphone let the engineman, the conductor, and the operator talk while the train kept moving. The Pennsylvania tested inductive gear from 1936, put a working plant on the Belvidere-Delaware Branch during the war, and strung it across the Middle and Pittsburgh divisions after 1945. General Order 2220 took it out of service on April 30, 1967.

Pennsylvania Railroad F7A diesel locomotives with Trainphone antennas running along the cab roofs

Pennsylvania F7As with the Trainphone bar along the cab roof. On a covered-wagon set the bar was a cab-unit fitting. American-Rails.com collection.

On this page: At a glance · Orders on the fly · The Bel-Del trial · The circuit · Harrisburg to Pittsburgh · Under the wires · April 30, 1967

At a glance

Item	Detail
System	Trainphone. An inductive carrier telephone. Union Switch & Signal, developed with the Pennsylvania Railroad
What the roof bar did	Transmitting loop. The field coupled into the rails or the pole-line wires. Useful range of the field itself was on the order of a hundred feet, which was enough to reach that conductor
Early experiment	Inductive work is dated from 1936. On March 4, 1926, a Fort Wayne Division freight had tried a telephone with an actual wire strung over the cars
Service trial	Belvidere-Delaware Branch, Trenton to Phillipsburg, about 50 miles, single track. Authorized September 1941. An engine and a cabin car were equipped by June 1942
First dated wayside station	Frenchtown, New Jersey, November 29, 1943, General Order 604
Trial carrier	5,700 cycles, amplitude modulation, fed into the rails
Production carriers	Two frequency-modulated channels. The low channel was 88 kHz. The high channel was 144 kHz on the sets adopted in 1944; a 1944 specification for Harrisburg–Pittsburgh gives 120 kHz for the second channel. The cab selector was marked L or H
Main-line plant	Harrisburg–Pittsburgh, 245 miles of four-track railroad. The 1944 order covered 275 locomotives, 90 cabooses, and 6 wayside stations. By 1946 the broader plant was about 300 locomotives, 100 cabin cars, 16 wayside stations, and 1,056 miles of main track
1950	1,613 miles reported in service
Wayside locations	210 inductive installations identified over the life of the system
Where it failed	Electrified track. Catenary current overwhelmed the inductive signal. The New York–Washington main and the Harrisburg extension never had a working Trainphone
Out of service	April 30, 1967, General Order 2220. VHF radio replaced it
Also used	The same Union Switch & Signal system on the Kansas City Southern and, on a smaller scale, the Duluth, Missabe & Iron Range. An earlier US&S installation was on the Bessemer & Lake Erie

A telephone instead of a hoop

Train orders moved on paper. The block operator copied them off the company wire, then got them to a moving crew with a hoop, or held the train until a brakeman could step off and take the tissue. Neither method was a conversation. The head end could not ask the cabin car a question. The cabin car could not tell the engineman that a brake was sticking until somebody walked the train or the train stopped.

Commercial radio was already a household appliance in the 1930s. The railroads did not have a block of frequencies of their own, and a transmitter strong enough to be useful brought a license with it. Induction stayed on the property. The field died out within a short distance of the rail and the pole line, which kept the circuit out of the Federal Communications Commission’s broadcast bands. Union Switch & Signal, the Pennsylvania’s signal supplier at Swissvale, had already built the railroad’s pulse-code cab signals on an inductive principle. The Trainphone was the voice version of that idea: a telephone whose wire, for the last hundred feet, was a magnetic field.

The Pennsylvania had tried the blunt version first. On March 4, 1926, an eastbound freight on the Fort Wayne Division ran with a telephone between engine and caboose and a wire laid over the tops of its 73 cars. It worked for that train. It was not a system. The inductive experiments that became the Trainphone are dated from 1936.

1941: the Belvidere-Delaware

The railroad authorized an experimental Union Switch & Signal installation on the Belvidere-Delaware Branch in September 1941. The Bel-Del was a good laboratory. It was about 50 miles of single track from Trenton to Phillipsburg, New Jersey, busy enough with freight to need head-to-rear talk, and free of catenary. By June 1942 one engine and one cabin car were equipped, the wayside gear was in, and the circuit was in service. The trial crew grew to ten engines, ten cabin cars, and a block station at Frenchtown, halfway along the branch, with a second station to follow.

That first carrier was a plain one. The set put voice and a calling signal on the upper sideband of a 5,700-cycle current, amplitude-modulated, and fed the current into the rails through the trucks. Coils under the vehicle picked the rail current back up. The same current induced a voltage in the communication wires on the adjacent pole line, and those wires, tied into the block station, carried the talk farther than the rail alone would. A freight could speak from the engine to the cabin car, and either end could raise Frenchtown, without a stop.

In February 1944 the Pennsylvania said the Bel-Del test had worked. Frenchtown’s inductive set, the first wayside station with a surviving general-order date, went into listed service on November 29, 1943. Later that year the railroad tried a different carrier on the same branch: two frequency-modulated channels, 88 and 144 kilocycles, in place of the 5,700-cycle AM set. The FM circuit was the one that went into production. On the cab control box a lever selected the low channel or the high one. Crews knew them as L and H.

Roof bar, handset, and the Carryphone

A production installation had three pieces. On the train, the roof tube was the transmitting loop, and a pair of receiving coils, sometimes the round disks on a cabin-car roof, took the incoming carrier off the rail or the wayside wire. In the cab, or in the cabin car, there was a telephone handset and a loudspeaker that called attention when the channel came up. Crews took the message on the handset. The speaker was there so a man with both hands on the brake or the fire would know somebody wanted him.

On the ground, the “shore” conductor was whatever already ran beside the track. The rails carried the carrier a short distance. The pole-line wires carried it to the next equipped block station, and the station equipment was wired to those conductors. Where the existing open wire sat too far from the rail to catch the field, the railroad strung a separate wire in closer and kept it clear of ground. There was no chain of loop antennas standing at junctions. The wayside plant was a tap on a wire the railroad already owned.

Two channels kept the party line usable. Everyone coupled to the same wire on the same channel could hear everyone else, which is what a head-to-rear circuit on a freight was supposed to do, and what a meet between two trains on the same block was supposed to do. The second channel gave the operator a place to call a train without sitting on top of that conversation. It was still a party line, and the audio was the audio of a carrier induced into a telegraph wire: usable, and nobody’s idea of a quiet circuit.

In 1946 the railroad added a portable piece and called it the Carryphone. It worked on the 88-kilocycle channel of the parent system. A conductor walking a train for a brake inspection, or a man on the ground at a crippled car, could talk to the engineman through the same inductive plant the engine and the cabin car were already using. The publicity photograph of the period is exactly that job: a freight conductor in the ballast, handset in hand, the train standing for an inspection he no longer had to finish before he could tell the head end what he had found.

Steam engines wore the bar on the tender, where a receiver coil and a loop could sit clear of the cab roof and the stack. Cabin cars wore a loop the length of the roof. On diesel consists the bar is the cab-unit signature in the photographs: F-units, Baldwin sharks, E-units, and then the hood units. A trailing booster in an F-unit set ordinarily had nothing on the roof to transmit. The lead unit and the cabin car were the two telephones on the train.

West of Harrisburg

The Bel-Del proved the circuit. The railroad then ordered it for the place the tonnage was. A 1944 account of the Union Switch & Signal system then going into production describes 245 miles of four-track main line between Harrisburg and Pittsburgh, with 275 locomotives, 90 cabooses, and six wayside stations on order. The two carriers in that specification are 88 and 120 kilocycles. By 1946, with the FM standard settled on 88 and 144 kilocycles, Railway Signaling counted about 300 locomotives, 100 cabin cars, and 16 wayside stations in regular service, on 1,056 miles of main track, Harrisburg–Pittsburgh plus the Bel-Del.

The spread after the war is a mileage story and a station story. By 1950 the figure in print was 1,613 miles. A count two years later put 1,268 Trainphone sets on the property. Chuck Heuer’s roster of inductive wayside stations, compiled from employee timetables, general orders, and the Pennsylvania Railroad Technical & Historical Society archives, finds 210 places that had the inductive set at some point in its life. The early names are the Middle Division and the Pittsburgh Division towers, beginning in 1946: Port, Lewis, Jacks, and the towers west of Altoona, then the Pittsburgh Line. The roof bar followed the freight. Passenger engines on the unelectrified divisions got it, and then the freight power and the cabin cars.

Pennsylvania Railroad RF16 sharks and an F7A-F3B set with Trainphone antennas at Erie Pennsylvania in October 1958

RF16 sharks and an F7A–F3B set at Erie, Pennsylvania, in October 1958. Every cab in the line has the roof bar. William Volkmer photo.

Other railroads bought the same idea, and some of them bought this system. Union Switch & Signal already had an inductive installation on the Bessemer & Lake Erie before the Pennsylvania’s main-line order. The Kansas City Southern adopted the Trainphone, and the Duluth, Missabe & Iron Range used it on a smaller scale. Officers of the Atlantic Coast Line came to look at the Bel-Del and ordered the system for 234 miles. The Milwaukee Road tested it. Parallel inductive telephones, from other makers, showed up in yards on the New York Central, the Louisville & Nashville, the Norfolk & Western, and the Burlington. The roof bar in a photograph is the Pennsylvania’s version. The problem it solved was everybody’s.

The one railroad it could not serve

The inductive signal was a weak current in a rail or a telegraph wire. A 11,000-volt catenary, and the traction current coming back through those same rails, buried it. The Trainphone did not work on electrified track. That was not a side issue. The Pennsylvania’s busiest passenger railroad was electrified: New York to Washington by February 10, 1935, and on to Harrisburg on January 15, 1938. The GG1 territory, the place a dispatcher would most have wanted a voice in the cab, was exactly the place the roof bar was silent. West of Harrisburg, and on branches like the Bel-Del, the circuit earned its keep. Under the wires, train orders stayed on paper.

The limitation also explains why the bar lasted as long as it did on freight power and disappeared from the story of the Northeast Corridor. A system that cannot see the railroad’s premier main line is a freight tool. The Pennsylvania kept buying it, and kept specifying it on new diesels, because west of the catenary it was the tool the company had paid Union Switch & Signal to build.

New Pennsylvania Railroad GP30s including 2222 with Trainphone antennas at Enola Pennsylvania in April 1963

New GP30s at Enola in April 1963, with #2222 nearest the camera. The railroad was still specifying the roof bar on road power four years before the system came out of service. Fred Byerly photo. American-Rails.com collection.

VHF, and General Order 2220

In May 1945 the Federal Communications Commission set aside a block of VHF channels for the railroads, between 152 and 162 megahertz. A licensed radio could leave the pole line behind. It worked under catenary. It reached a tower that had no inductive tap. It did not have to induce a whisper into a telegraph wire and hope the next station could dig the voice back out. Railroads that had never bought an inductive plant started installing VHF sets by the thousand. The Pennsylvania, which had, did not. Through the 1950s its radio purchases were the yard jobs, Sunnyside among them. The freight trains on the Middle Division still wore the bar.

The word drifted. After 1955, employee timetables on the Philadelphia Region began to call some new yard and terminal installations “trainphone” when the hardware was VHF. Heuer’s roster leaves those out, about ninety of them from 1957 on, because they were never part of the inductive system. A timetable note from that decade has to be read against the territory. West of Harrisburg, Trainphone still meant the roof bar. Around Philadelphia, the same word was starting to mean a radio.

The inductive plant itself lasted until the end of the Pennsylvania. General Order 2220, effective April 30, 1967, took the Trainphones out of service. Ten months later the company was gone into Penn Central. The bars stayed on some cabs after the sets behind them were dead, which is why a 1966 or 1967 photograph can still show the tube. The GP30s at Enola in April 1963 are the late specification: second-generation road power, delivered with a telephone the railroad had been building since the Bel-Del, on a railroad that had already been offered a better one for twenty years.

What replaced the handset was ordinary railroad radio, the VHF set in the cab that a crew still uses. The roof bar remains the easy thing to see in a Pennsylvania photograph from 1945 to 1967, and the reason a model of an F-unit, a shark, a GP9, or a GP30 from those years is unfinished without it.

Sources

Union Switch & Signal and the Pennsylvania Railroad, inductive train-telephone development, reported in Railway Signaling: the 1944 account of the production system (88 and 120 kilocycle channels; 245 miles, Harrisburg–Pittsburgh; 275 locomotives, 90 cabooses, 6 wayside stations) and the 1946 Carryphone article (5,700-cycle AM trial replaced by 88 and 144 kilocycle FM; about 300 locomotives, 100 cabin cars, 16 wayside stations, 1,056 miles of main track).
Belvidere-Delaware trial: authorized September 1941; an engine and a cabin car in service by June 1942; 5,700-cycle amplitude-modulated carrier fed into the rails. Contemporary technical account of the Union train-communication system. Frenchtown wayside station, November 29, 1943, General Order 604.
Newman, M. S. “Train Talk.” Railroad Magazine, October 1944. The Bel-Del announcement of February 1944.
“Voices on the Rails.” Trains, August 1997. The 1,613-mile figure for 1950, the Atlantic Coast Line’s 234-mile order, and General Order 2220, effective April 30, 1967.
Heuer, Chuck. “Roster of Inductive Trainphone Locations on the PRR.” December 2022. Two hundred ten inductive wayside locations; the 1957 Philadelphia Region notes that were VHF yard radio rather than induction.
Pennsylvania Railroad chronology, March 4, 1926: telephone trial on a Fort Wayne Division freight, with a wire strung over the cars.

Related pages: Pennsylvania Railroad · PRR diesel roster · GG1 · Pole lines · Bessemer & Lake Erie

Facebook X Pinterest Tumblr Reddit WhatsApp Messenger