Berkshires and Kanawhas
by Edward G. Weinstein

Boston & Main 4017
The introduction of the 2-8-4 wheel arrangement by the Lima Locomotive Works in 1925 is generally accepted as the commencement of the "Super Power" era of the North American steam locomotive. The term generally has been used to denote large road locomotives with relatively high drivers, large fireboxes and relatively high working pressures. One characteristic generally considered to be essential was a four wheel, and, in some designs, six wheel. trailing truck. (This latter characteristic, to my mind, is misleading: I can think of at least one class of 4-6-2s and three of 4-8-2s which, on their service achievements, were as much "Super Power" as they would have been if built as 4-6-4s or 4-8-4s.) If a reader has access, as your author has, to old editions of the "Locomotive Cyclopedia", using the 1930 edition as an example one sees that the wording used is "Lima Super Power". But, the reader will also note that this wording appears only in paid advertisements for the Lima Locomotive Works. Unfortunately, to my mind, while the term "Super Power" is a perfectly fine one, there seems to have been a tendency among some to assume that all locomotives falling into the category were either designed by or followed design principles first enunciated by the Lima Locomotive Works. (This can result in some oddities, such as the title of a fine book in my library. "Lackawanna: Superpower Railroad of the North East", about a railroad which never owned one locomotive built or designed by Lima.)

The key event in the development of the steam locomotive designs which brought North American railroads into 1950 was the introduction, about 1910, of a successful method of delivering high temperature superheated steam to the cylinders. (This was the "Schmidt" superheater, designed by Wilhelm Schmidt, 1859 - 1924. Schmidt was a German, generally associated with the Prussian State Railways, although the initial installation of his definitive superheater was in Belgium in 1901. In North America, the Locomotive Superheater Corporation, generally known as Elesco manufactured his device.)

To illustrate the importance of the superheater, one authority (Bruce) stated that, for the same amount of fuel burnt, a superheated locomotive would produce 25 to 30% more power than one of the same specifications but retaining saturated steam. Although its introduction was not without problems, the superheater was so vital an appliance that, by 1912, for all practical purposes, all road locomotives were being built superheated, and many roads were extensively retrofitting superheaters to existing locomotives. (Bruce)

Boston & Main 4021
By 1900, most simple locomotives were being built with boiler pressures in the 190/200- psi range. (Compounds generally had higher pressures, but the introduction of superheating effectively ended the production of rigid frame compounds, and, indeed, resulted in the rebuilding of many such as superheated simples. The concept of superheated compounds, so important in France as well as some other European countries, never was attractive to North American roads.) The introduction of superheating, in many cases, caused a reduction in working pressures - Great Northern 2-8-2s of 1911 and Santa Fe 2-10-2s of 1912 with 170-psi boilers. The reasoning here, not, perhaps, among the engineers and draftsmen who were the actual designers, but among the operating and financial types who told them what to build, was that, undeniably, reducing boiler pressure was a savings in both fuel consumption and maintenance costs. If reduced boiler pressures resulted in lower starting tractive efforts, the answer would be larger cylinders, or, less frequently (because of counterbalancing problems) lower drivers. If, as a result, trains, freights in particular, had to operate more slowly because the reduction in boiler pressure without a concomitant reduction in cylinder volume would result in a speed-tractive effort curve which would vanish at about 30mph, that was an acceptable result. (Let it be understood that not all railroads followed this particular motive power policy. The Pennsylvania, for one, was consistent in ordering new locomotives where the working pressure was a respectable 205-psi, and in 1916, developing a new design with 250psi. although this pressure was required to permit the use of a cutoff limited to 50% of full stroke.)

In any event, by 1916 the railroads had come to realize that there was a requirement for speedier freight service. Partially, this was due to the realization that there were potential cargoes, which might be developed with better service. Perhaps it was because the opening of the Panama Canal meant that a ship could leave New York, make stops on the way, and be in San Francisco in about 21 days, and offer much better rates than could a railroad, it behooved the roads to try to get their service moving a bit faster. And, assuredly, after 1918, there was a distinct perception that, at some point motor freight could be a competitor, at least for high value lightweight commodities.

Boston & Albany 1411
So, the USRA locomotives, as a group, were much better designed, on most roads, which received them, than were the roads own pre-1916 designs. They had one characteristic, which was basic to "Super Power"; these locomotives had boilers, which had ample capacity. Thus, in a way, the assignment which Lima had from the New York Central was to upgrade its 1912 H-7 class, (n the H7E group, the last ten built in 1920, had had their boiler pressure increased from 190 to 200 psi) to something at least as good as, but, hopefully, better than the USRA Mikados. (The NYC had received 184 " light" and 30 "heavy" USRA Mikados, road classes H-6 and H-9, so they knew what these locomotives could do.)

Some years ago in "Trains" there appeared an article in which the author addressed the question of what was the reason why the Lima designed H-10 class turned out to be such an outstanding design. The conclusion reached in the article was that the secret was in the superheating. The H-10 had 1700 square feet of superheater surface in contrast to the 1163 square feet of the H-7E, and the Lima design incorporated the type "E" superheater, superior to that equipping the H-7. There were other changes in basic specifications; the H-10 had slightly larger grate area, and 28 x 30 cylinders instead of 27 x 30. (Actually, this was not an improvement) The 'appliances" - booster, feed water heater and the like, were extraneous to the superheater increase. (All of these developments are covered brilliantly, and in great detail, by Eric Hirsimaki in his 1986 book, "Lima - the history.") It seems clear that the next step, the prototype Berkshire was built with some sort of commitment from the New York Central. The design was basically an enlargement, physically, of the H-10, in so far as clearances, weights and similar items were involved. However, there was an odd situation. We have mentioned that, by and large, the railroads had, after the general introduction of superheating, been willing to install locomotives best suited for slow freight operation. We have also seen how the Pennsylvania did not follow this regressive design policy. The New York Central also, in a different fashion, followed motive power policies which provided for fast freight service. In 1910, they purchased the first in a large series of 69 inch drivered Pacifics, intended for fast freight service. (Other roads, the Erie and Lackawanna for two, ordered similar freight 4-6-2s). And, in 1916 - 18, the NYC installed 185 69-inch drivered freight 4-8-2s. (This was an amazing move; 4-8-2s were conceived as heavy, or "Mountain" passenger engines, and were so regarded by the USRA.) The NYC had an extremely complex system of speed limits, with all sorts of caveats concerning tonnage, train length, type of locomotive, the actual tracks on which trains were operating. But, one absolute rule was this: in either passenger or freight service, no locomotive which did not have a four wheel leading truck, and drivers of at least 69 inch diameter, was permitted to run at more than 50 mph. In freight service, where the speed limit in 1925 was 60mph, this meant that Limas A1A was going to be assigned to drag freight service, or, as it worked out, assigned to one of the relatively few parts of the New York Central system where main line grades were a problem, the Boston & Albany.

Boston & Albany 1447
The best comparison which shows the advantage of Lima's new design over the existing power is the classic one of the test run eastbound from Selkirk, in which A1A overhauled a train hauled by H-10 #190, averaging about 31 mph against the Mikado's 21 mph over the 46 miles involved. Note also that the Lima engine was hauling about 2300 tons, against the roughly 1700 tons of 190. So, on a gradient profile, which, in places, was about 1% opposing, the A1A handled a heavier train much faster than 190. (Please be aware that this was not a matter of higher top speed; the B&A was a 45mph railroad for freight trains hauled by Mikados or Berkshires. The difference was that the A1A was more powerful in the higher speed ranges than the H10. (Note that the on paper difference in starting tractive effort is meaningless; the weight on drivers of both locomotives was the same, and both were equipped with boosters.)

The success of A1A on the B&A, and on its other road tests, lead to a substantial amount of orders for Lima: in 1926-30 there were 55 units ordered for the B&A; 49, plus A1A itself, for the Illinois Central; 25 for the B&M. 25 for the MP. This was a substantial addition to the order book of what was the smallest of the major locomotive builders in the United States. Other builders participated; Alco built 5 for the MoPac's Texas subsidiary, the International Great Northern, and 2 for the Toronto Hamilton & Buffalo, built by Alco's Canadian manufacturing subsidiary, Montreal Locomotive. These were identical in basic specifications to the Lima design. Alco also built 12 similar 2-8-4s for the Chicago & North Western. The AT&SF ordered 15 2-8-4s from Baldwin, differing not only in the omission of limited cutoff but in having a more modern cylinder dimension, 27 x 32, and a more realistic working pressure of 220 psi. Limited cutoff was a concept, best exemplified by the Pennsylvania's fleet of 2-10-0s, I-1s and sub classes. These locomotives had valve gear designed to admit steam to the cylinders during only 50% of the stroke, as opposed to the more normal 85%. The concept here was that this would enable steam to expand fully, thus resulting in greater fuel economy. To make this concept work, increased steam pressure was required to achieve the desired starting tractive effort. Thus, PRR 2-10-2s of about the same vintage as the I-1s had 205 psi, while the Decapods had 250 psi. The Lima 2-8-4s had 60% cutoff, and, as originally designed, 230 psi. This created marketing problem; the initial assignment of the Berkshires on the B&A was to replace USRA light 2-10-2s and 2-6-6-2 Mallets in the district between Selkirk and Springfield, MA. The 2-10-2s, class Z-1, had a starting tractive effort of about 68,000 lbs, the Mallets, 63,000 working compound. The new locomotives had to show an at least an equal starting to those they replaced. Thus, the A1A was built with 240 psi.

Erie 3316
In the event, this limited cutoff feature was removed from the B&A engines, so that by the end of 1945, all were working with the normal 85% cutoff and 220 psi. The additional fuel required to achieve the higher boiler pressure, plus the additional maintenance costs of higher pressure outweighed and fuel savings. In the particular case of the Berkshires, there was a further complication; these locomotives had a poor factor of adhesion; the limited cutoff resulted in very uneven power when starting, thus making a bad situation worse.

In all fairness, I should point out that limited cutoff was a feature of many highly successful designs; thus, what in my mind were the two best North American heavy fast freight locomotives, the Santa Fe 5001/5011 classes of 2-10-4, and the Norfolk & Western class A 2-6-6-4 had limited cutoffs of 67% and 75% respectively. These modest limits were, I believe, used to ensure acceptable factors of adhesion when starting without limiting the boiler pressure required to achieve high horsepower output at high freight train speeds.

However, the limited cutoff feature accentuated a severe drawback to the A1A design; the higher boiler pressure required to accommodate limited cutoff, plus the overly large cylinder dimension inherited from the H10, resulted in extremely high piston thrust, which was difficult to counterbalance adequately with a 63 inch diameter driver. It is interesting to note, that, at the point when A1A locomotives were operating with the boiler pressure having been reduced to 220 psi with the removal of limited cutoff, they were still limited to 45 mph on the Boston & Albany, while 69 inch drivered Mohawks were permitted a freight train speed of 50 mph. Even more indicative of the Berkshires' capability for track damage was that, on the River Division (West Shore), where the A1A were limited to 40 mph, while all other types of 63 inch drivered freight power could operate at 50 mph. Another omission on the A1A was that off a combustion chamber; this feature, which became part of just about every large road steam locomotive built from the mid 1930s on, was regarded with disfavor by the New York Central, on the grounds of higher maintenance costs. (Although at the same time as the A1A class, and the J1 Hudsons, for that matter, were being designed without combustion chambers, Alco was designing the first of 300 L2 Mohawks, with combustion chambers.)

Erie 3360
While there were earlier locomotives built with four wheel trailer trucks, these were all of an experimental nature. Lima was the first to realize that the requirement for larger grate areas would require the use of four wheel trucks, as trailing axle loads on some Mikados and Santa Fe's were approaching, and in some cases exceeding 60,000 pounds. On the other hand, the engineering required to enlarge the existing single axle designs to two and. later three axles was not substantial. (Of course, Lima's own concept of a trailing truck serving as a link for transmitting pull between the drivers and the tender, and having the ash pan built as an integral part of the truck, was unsuccessful.)

However, while the B&M and the B&A were still receiving 63 inch drivered "Superpower" Berkshires, something different was being built at Alco's Brooks works. This was a 70-inch drivered 2-8-4 being built for the Erie. Ultimately, there were 105 of these locomotives built between 1927 and 1929; 25 from Alco, 25 from Lima, 35 from Baldwin and then the final 20 from Lima. Erie personnel did not design these locomotives, classes S1/4, nor did the builders design them. Instead, a committee designed them.

Pere Marquette 1206
The Erie in 1927 was part of a group of railroads known as the "Van Sweringen Lines", after the two brothers from Cleveland who controlled the Erie, as well as the Chesapeake & Ohio, the Pere Marquette, and the New York Chicago and St.Louis, more generally known as the "Nickel Plate." The Mechanical Departments of these roads formed the "Advisory Mechanical Committee", which served as the locomotive (and, one assumes, car) designers for the group. The Erie was in the difficult position of having to compete for overhead freight traffic between New York and Chicago with both the New York Central and the Pennsylvania. To do this successfully, the Erie needed something new and different in freight locomotives.

What the Advisory Mechanical Committee designed was a giant 2-8-4, with 70-inch drivers instead of the 63 inch ones of the A1A and the existing Erie Mikados. The significance of higher drive wheels is three fold; for one thing, the higher the driver, the lower the piston speed in relation to track speed. Thus, the 70-inch Erie drivers had a piston speed about 35% less than the 63-inch drivers of A1A. It is a given that, at equal track speeds, the lower the piston speeds the higher is the mean effective pressure in the cylinders, all other factors being equal, obviously. Secondly, the lower rotational speeds are beneficial to maintenance. Finally, the higher the driver diameter, the more area is available for counterbalancing.

A1A had a large boiler; the Erie Berkshires had a huge one. Grate area was identical, 100 square feet; but the Erie (my statistics here are based on Alco built 3300-3324) engines had 5669 square feet of evaporative heating service against the 5210 square feet of A1A, and the Erie superheating surface was 2448 square feet as opposed to 2211 square feet. It is important to note that the impressive figures of the Erie design were related to the less restrictive clearances and greater weight on drivers permitted by the Erie over the restrictions imposed on Lima by the necessity to design to fit NYC requirements. Thus, A1A had an outside boiler diameter of 94 inches, while the Erie engines had 100 inch boilers. The weight on drivers for the S1 was 276,000 lbs, that of A1A 246,200.The Erie did not employ a combustion chamber.

New York, Chicago & St. Louis 708
The four classes were identical in driver diameter. They all had 70 inch drivers, and the cylinder dimensions were 28 1/2 by 32, more adaptable to long lap/long valve travel operating practices than the 28 x 30 cylinders of A1A. All 105 had boosters. One difference was this; the 25 Alco and 35 Baldwin built locomotives had 225 psi working pressures and normal cutoff; the 55 Limas, 250 psi and 60% cutoff, resulting in slightly higher starting tractive effort and slightly less adequate factor of adhesion. The Alcos had a conventional Delta trailing truck, but the Baldwins and Limas used the same sort of articulated trailing truck as A1A (There were differences among the tenders of the four classes.).

These were extremely successful locomotives. With the exception of one major improvement, they remained unaltered until they began to leave the roster in the early 1950s. This was the rebuilding of most, if not all, of the Berkshires with cast one-piece underframes, including the cylinders, on the front half of the chassis. This was a major improvement in increasing availability and reducing maintenance costs.

From the Erie Berkshire, the Advisory Mechanical Committee developed the Chesapeake & Ohio class T-1 2-10-4. This design was based on a test of Erie 3377, a Baldwin, which for test purposes had its boiler pressure increased to 250psi and with additional weight on drivers. The T-1 design had 265psi, a combustion chamber, 69-inch drivers and 29 x 34 cylinders. (Originally, the locomotives came from Lima with 260 psi and 60% cutoff; the boiler pressure was increased, as above, and normal cutoff was used.)

The subsequent development history of the 2-8-4s is relatively simple. Starting on the NKP in 1934, and on the Pere Marquette in 1937, two related types of 2-8-4 were built. The earliest, for the Nickel Plate, was the famous class S. This was a modified T-1, with the same 69-inch drivers, 245 psi and 25 x 34 cylinders. This design was replicated many times on the NKP, and locomotives with the same basic specifications were purchased by the Wheeling & Lake Erie, which, ultimately, wound up on the NKP roster when the W&LE merged with the Nickel Plate. Under wartime circumstances, ten identical locomotives were built for the Richmond, Fredericksburg and Potomac, a line, which had relied on 77 inch drivered Baldwin 4-8-4s for its freight and passenger trains.

Wheeling & Lake Erie 6404


Chesapeake & Ohio 2754
Beginning in 1937 on the Pere Marquette, a variant 2-8-4 was developed, which, ultimately, lead to the C&O type known by that road as "Kanawhas". These were identical with the NKP/W&LE except for having larger cylinders, 26 x 34, and more weight on drivers, so they exerted about 5000 lbs more starting tractive effort. (While not at all slow by freight standards, these locomotives with their bigger cylinders and greater weight were less speedy than their NKP cousins.) Oddly, these larger 2-8-4s had slightly less superheater service than their smaller ancestors. Starting in 1944, the C&O began ordering locomotives basically identical to those of the Pere Marquette, and, of course, the Pere Marquette locomotives, through a merger, ultimately became part of the C&O roster. 5 identical locomotives went on the roster of the Virginian in 1946.


There was one final group of 2-8-4s built which some consider "Van Sweringen" engines. These were built for the Louisville & Nashville, initially 14 locomotives by Baldwin in 1942, followed by 6 more in 1944 and a final 22 from Lima in 1949. The had two design differences from the NKP pattern; they had 25 x 32 inch cylinders, but, because they operated with a maximum steam pressure of 265 psi had about 1000 lbs greater starting te. There were other differences, but also many similarities to the Van Sweringen designs. Given the fact that, by the 1930s, large steam locomotives were constantly running into weight and clearance restrictions, the amount of variety which h a designer could put into a design of a given wheel arrangement was highly limited, one can accept or reject the imitation of the Van Sweringen pattern argument as one pleases.

Virginian 505
The Berkshires built after 1934 had all of the characteristics, which made up "modern" steam; large grates, combustion chambers, large boilers and large superheater surfaces typified the steam generating plant. Cast one-piece engine beds simplified maintenance, as did the application of roller bearings to a greater or lesser degree. Boosters were applied in many cases. The lack of universality of what had been one of the basic items of the initial Lima "Superpower" specifications was because of the high maintenance costs of boosters. Thus, those roads, such as the Louisville & Nashville, which expected that their Berkshires would be spending much of their time in mineral traffic, would always be equipped with boosters, even if the locomotives would be occasionally used in fast freight or even passenger service. On the other hand, the RF&P, a fast freight route, did not fit boosters; the additional tonnage which could be started with a booster could not be hauled at the track speed which the road's schedules demanded.

There were three other classes of 2-8-4 built after 1934, which are included in the total of 611 North American Berkshires, starting in 1925 with A1A and ending with NKP 779 in 1949. (Incidentally, it should be remembered that the largest single group of 2-8-4s was the 646 "IS", for "Iosif Stalin" class of the former Soviet Union. These were much smaller than the run of the North American types above, and had 71-inch drivers for passenger service. In fact, in Europe all 2-8-4s were passenger engines; in Austria, Romania, Czechoslovakia and Norway, as well as Russia.) These were, in order of construction, 4 built by Lima in 1934, followed by 2 more in 1939, for the Detroit, Toledo and Ironton, a north-south line which ran against the grain of the country on its main line. These were about the same weight as the NKP/W&LE engines, but had a higher boiler pressure, 260psi, 63 inch drivers, but small 25 x 30 cylinders, thus keeping piston thrust within reasonable levels, They had almost 2000 lbs more starting tractive effort than the Van Sweringen Berkshires, and, more than A1A and its siblings, was a "Super Mikado". The next was the smallest 2-8-4, 5 units built in 1940 for the then Norfolk Southern. (This line subsequently became part of the Southern Railway, thus today being a component of todays Norfolk Southern) These were small locomotives, 334,000 lbs total engine weight, roughly 75 tons less than an Erie 2-8-4. They had 63-inch drivers, 23 1/2 x 30-inch cylinders (60% cutoff) and 250psi boilers. They had a starting tractive effort of 49,900 lbs, plus 11,000lbs from the booster.

The final design was the New York Central (Pittsburgh & Lake Erie) class A2a, built by Alco in 1948, and that builders last steam locomotive order. There were seven of these locomotives. To my mind, they represent Paul Keifer's concept of what a general purpose stream freight locomotive should have been. For the first time, the NYC permitted a 70,000lb axle loading. Thus, with good adhesion, this class had a starting te of 67,000. Since they would have been confined to the 50mph and under freights, they could have made use of a booster, which would have given them a starting te of about 80/81,000 lbs. Even without the booster, they were rated at 8000 tons on the admittedly water level route of the P&LE. They had a modest steam pressure of 230psi, but a boiler capacity equal to that of a NKP 2-8-4, and 25 x 32 cylinders.


Much thanks to Edward G. Weinstein who has provided all of the information found on this page. All photos are from Edward G. Weinstein's collection.
All material Copyright ©
Wes Barris