Thomson Rail Welding 1910

Impartial Scientific Review of Electric Welding of Rail Tracks Published in 1910

The Thomson rail welding machine was used with effect from February 1893 for laying 3½ miles (5.6 km) of rail at the Baden & St. Louis Railroad and in 1917 for "Compression Welding" of rail track. The following impartial scientific review was published by Richard Newell Hart in 1910 in his book on Welding Theory, Practice, Apparatus and Tests:


Rail Welding by the Thomson Process

By Richard Newell Hart, B.S. (* 1882)

Published in: Welding Theory, Practice, Apparatus and Tests

Electric, Thermit and Hot-Flame Processes

McGraw-Hill Book Company, New York, 1910[1]

The Thomson Process

This process differs radically from all the others in forcing through the metal to be heated electrically such volumes of current that its own resistance is sufficient to bring every molecule of the section traversed by the current to the desired temperature.[2] Current is taken from a lighting or power circuit, is stepped down to the required 3 or more volts and higher volume, and is passed through a secondary circuit in which the greatest resistance is offered by the pieces of the metal to be welded. The cross-section and unit resistivity are so proportioned to the flow of current that the resistance produces red or white heat at the point of welding. The hot metals are then forced together and the weld is made. The apparatus necessary are:

1. A generator of alternating current.
2. A step-down transformer, carried in the body of the welder.
3. Apparatus for regulating the current, and sometimes
apparatus for automatically shutting off the current as soon as
welding heat is reached.
4. Clamps for holding the metal to be welded and to transmit
the current to it.


The Thomson process presents a number of decided advantages. Among them:

1. It is at present the best all-around welding machine for welding continuous runs of one weld, such as printers' chases.
2. The power used is claimed to give a 75 per cent, heat efficiency; the power is used only as long as needed, and is turned off as readily as the hot-flame welding burners.
3. The heating is rapid, even, entirely local, and is under control.
4. There is no excessive heating as with the electric arc; hence no excessive oxidation or decarbonizing of the metal.
5. The clamps hold the work in accurate alignment and furnish pressure enough to squeeze well the hot metal.
6. The workman is in no danger of injuring his eyes by excessive light, nor is the current at all dangerous. The operator works without dark glasses or protective apron and can hold the metal bars while the welding is going on.


The present limitations of the process seem to be:

1. Though it will weld odd or job work, it is practically limited to continuous welding of one article, known as repeat welding.
2. Though such metals as brass and cast iron can be welded on the Thomson machine, the company does not recommend it for such metals as have a marked melting point and which are not plastic below that point. High-carbon steel does not give an altogether satisfactory weld with this process.
3. The machine demands power at irregular intervals. For this reason station engineers may object to having a single machine of large size on their lines.

Apparatus and Current

Please refer to pages 43-66 of the original book for a detailed description of the apparatus and current.


Rail Welding by the Thomson Process

The most important single application of the Thomson process has been to the welding of street-car rails. Before 1892, all rail welding was done by the cast-welding process (…).
Recently the electric roads have begun to adopt the electrically welded rail and also the thermit-welded rail. Welded rails are a great improvement over those joined by fishplates and bonded with copper wire, for conducting the current: 

1. The conductivity of the weld is as good or better than the unit section of rail. There is no bonding to come loose or leak or be stolen.
2. The rail will last much longer.
3. Welded tracks is smoother riding.

Rails running through city streets are well embedded in the street. If the street paving is not a good conductor of heat and the extremes of summer and winter temperature are not too great, very long sections of track can be welded into one piece without fear of pulling loose at the ends or at any of the joints. A section of 2300 feet has been solidly joined at Holyoke, Mass. It is calculated that the coefficient of expansion of steel in such a climate would cause a stress of about 16,000 pounds to the inch (110 N/mm²), while the tensile strength of the rail would run well over 40,000 pounds (275 N/mm²).
Friction of the pavement against the rail and inertia of the rail prevent dragging, and the expansion and contraction are taken up by the elasticity of the rail. Rails welded with thermit or by electricity are less liable to crack or pull apart at the weld than are cast-welded rails.
The Thomson process was the first process of welding applied to the production of continuous rails on electric railway tracks, and was introduced by the Johnson Company in 1892.
In 1897, the Lorain Steel Company, successors to the Johnson Company, improved the process and placed it actively on the market. Since that time it has been made use of in almost all the large cities of the United States, and the company found it necessary to double its equipment for this kind of work.


The joint consists of two bars welded to the web of the rail, one on each side. Three welds are made between the bars and the rail, one directly over the ends of the two rails and at each end of the bars. The central weld is made first. In cooling, the contraction of the bars draws the abutting rails together so that no opening remains across the head of the rail. 

Thomson special machine for welding rails in streets
Fig. 32. — Thomson special machine for welding rails in streets

The apparatus is mounted on four trolley cars, propelled by their own motors. The first car carries a sand-blast apparatus for cleaning the rails and bars.


The welder is suspended from a crane projecting from the front of the second car (see Fig. 32). The welder itself consists of a "step-down transformer for supplying current for heating the weld, and hydraulic pressure apparatus for supplying a heavy pressure to the portions to be welded." Suitable mechanism is carried within the car for raising and lowering the welder and to swing it from side to side to engage either rail.

Coupled to the welder car, the third car carries rotary transformer and regulating apparatus for changing the direct current from the trolley to alternating current. A switchboard with instruments, etc., is also carried in this car.
The fourth car carries two grinder carriages, one suspended over each rail, to smooth down any inequalities that may exist on the head of the rail after the joint has been welded and to produce a true running surface.
The process has been successfully applied to all kinds of rail, both girder and T-rails. Also to the welding of the "third "or conductor rail on elevated and surface lines.
The process particularly commends itself for use in crowded city streets on account of its harmlessness, as it is not affected by dampness and there is no danger of explosions, etc., due to sudden rain storms. The apparatus is practically noiseless in its operation.
An interesting application was the welding of the T-rail on the surface track on the north and south roadways of the Brooklyn Bridge in 1906.
The cost of the equipment makes it more desirable for a railway company to have the welding done for them than to do it themselves.
The apparatus is also made use of for welding heavy copper cables to the rails, either for overhead return or around special work. As the conductivity of the welded joint is greater than the rail, a most perfect system of bonding is thus afforded at the same time with the elimination of the joints.
From ten to twenty welds are made per day by this machine. The breakage is said to run less than 5 per cent., and often not higher than i per cent. The machines are leased, not sold, and the cost must accordingly be figured on the rental, power, and labor in calculating the cost per joint.

Electric Resistance Heater

Besides its use as a welder, the machine may be used as a preheater of metals to be brazed or bent. It will sometimes be preferable to braze or solder a joint, when the two metals cannot be allowed to lose their shape or have any of their substance pressed into an upset: the welder can then be used as a preheater. The current would be regulated to bring the metals to a slightly lower-than-welding heat and keep them at this heat. In brazing brass, this is the best- known method of preheating, because a torch preheater always burns out some of the zinc in the brass and oxidizes the copper.
The Thomson welder may be used to anneal spots in armor plate. This is done by connecting the positive to the armor plate and pressing the negative clamp against the spot to be annealed.


In general, tests of electric welds show that from 75 to 95 per cent, of the original strength of the metal is reached. In cases where the upset is not cut off, the strength can be in- creased above 100 per cent. Welds of low-carbon steel and low sulphur-and-silicon iron, if well made and worked or drawn after working, will approximate 100 per cent, in strength.
It is sometimes asked if the electric current does not damage the metal. Electric welding is no more harmful to the metal than any other process. In fact, the control of the heat is so exact and overheating and reheating so seldom happen, that electric welds run uniformly high in tensile and elastic strength. A "burned" weld seldom occurs — the oxid at the joint is forced out into the upset and ground off. It may be emphatically stated that the electric-resistance welds are the best yet made. As an instance, such a misused and overstrained utensil as a printers chase seldom gives at the weld.
Sir Frederick Bramwell[3] states that 1 1/8-inch (28.5mm) round bars can be welded in 2 1/4 minutes with an average tensile strength of 91.9 per cent, against four minutes' time and 89.8 per cent, strength when smith-welded.
The results of a series of tests of electrically welded metals carried on at the Watertown Arsenal[4] may be abridged as follows:
  • Twenty-nine broke at the weld.   
  • Seventeen within 2 inches (50.8 mm) of the weld.    
  • Eleven within the range of moderate heat.   
  • Two near the grips.    
  • Welds of wrought iron were 5 to 10 per cent, below unit strength; fracture fibrous or slightly spongy.    
  • Welds of steel were from 50 to 80 per cent, less than unit strength.    
  • Copper welded at 5 to 10 per cent, less than unit strength.   
  • Steel welded to wrought iron at about the strength of the iron.   
  • Brass gave an uncertain weld with wrought iron and had a strength at the weld of 8 1/2 to 16 1/2 tons to the inch (i.e. 130 to 250 N/mm²).   
  • Steel welded with German silver with a strength of 20 tons to the inch (i.e. 304 N/mm²).   
  • Some welds of steel were about unit strength and some of iron were above unit strength.   
  • A number of these bars had upsets, however, and the upset does not seem to have increased the strength very much. 
When electric welding was first tried out there was serious complaint that the welds were burnt, spongy, and weak. This was due to the fact that the metals were melted together and were not worked. The welding machines with automatic swage blocks prevent crystallization at the weld, as does also hammering after welding. The weld is still liable to be weak on the edge of the heating radius. Many joints that will hold at the weld will break an inch either side because the heat has destroyed the properties of the metal.

Remarks and References

  1. Richard Newell Hart (* 1882): Welding: Theory, practice, apparatus and tests. Electric, thermit and hot-flame processes. McGraw-Hill Book Company, New York, 1910, p. 42 and 66-71.
  2. Hermann Lemp: The Engineering Magazine, August 1894.
  3. "Electric Engineering Formula," p. 673.
  4. Transactions of the American Society of Mechanical Engineers, 1889, p. 97.