Elegy for Sturmey-Archer
Vernon Forbes of Columbia, Missouri
on what went wrong during S-A’s British years
I felt a great sense of loss when I heard of Sturmey-Archer’s closing in 2000. I remember sitting for a long time after I heard the news, feeling numb. I had sold, repaired, championed and ridden Sturmey-Archer gears for 21 years. The first shop I worked at was a Raleigh shop. They hired me because I could overhaul Sturmey-Archer hubs. They had all the many internal parts for several models of Sturmey-Archer gears in a wall of 5″x 6″ metal drawers. Small parts were in a case of 1″ x 2″ plastic drawers. Such a vast collection of ancient artifacts bore mute testament to a long and fine tradition of strength and durability that stretched to the beginning of the last century. I used to say the popularity of derailleur gears was little more than a fad in efficiency.
Derailleur bikes then were relatively flimsy and in need of constant maintenance to keep them working. Ten speeds were designed after Tour de France bikes. If cars were designed the same way they would all look like Formula One racing cars. Derailleur bikes, like their Tour de France counterparts, were not user friendly. Cyclists then had to switch between touch and sound modalities to “find” their gears. Even the most expensive derailleurs required careful installation and bicycle with a straight frame and chainline. Mechanics would carefully position and skilfully bend the front derailleur cage with pliers. Even then it could not handle more than a relatively narrow range of high gears in the hands of someone who knew how to perform what Frank Berto described as the “overshift-and-correct” shifting drill. In writing the history of derailleur chain gearing (Berto, et al., 2000), noted author Frank Berto described the remarkable craftsmanship of a Campagnolo derailleur saying “it will shift lousy forever”.
The metaphor of a derailleur bike being like a high-strung high-performance race car pushing the envelope of technology was viable then. Bicycles were like radio in the forties, recorded sound in the fifties or cameras in the sixties. Bicycle mechanics were like microscope repairmen.
By comparison, Sturmey-Archer (S-A) gears were user friendly and virtually maintenance free. From among the hurley-burley of designs and compatibility problems Sturmey-Archer emerged as a golden standard. Its design rarely changed and parts had always been available. The most desirable feature of S-A gears was ease of shifting. Instead of shifting while steering with one hand you just flicked a trigger on the handlebars. S-A gears were safe, simple and reliable.
This reputation eroded as Suntour and Shimano began making derailleurs so reliable you didn’t need repair parts. Power shifters, bar-cons, and self-adjusting front derailleurs began closing in on the superior simplicity of use previously dominated by S-A. In 1985 Shimano optimized rear derailleur geometry and introduced indexed shifting (Berto, et al., p.260). The final blow came in 1989 with SRAM’s twist grip for derailleurs (Berto, et al., 2000, p. 263) just like those S-A had been producing since 1961. Suddenly, all I could say to potential customers was that S-A gears were indexed shifting when indexed shifting wasn’t cool.
Losing their advantage of being both user friendly and low-maintenance in the face of a rapidly improved derailleur a fair comparison could now be made. Having a weak engine, bicycles need higher gears that are closer together and lower gears that are further apart. Hub gears have higher gears that are further apart and lower gears that are closer together (see Figure 1 below); the exact opposite of what is needed (van der Plas, p. 161). This is just how hub gears work. Derailleur gears, with their ability to match the limitations of human effort, were superior in this regard.
At least in America, few people seemed to know that the sprocket size on an S-A hub could be changed so that a larger sprocket with more teeth gave lower gears. The gears Raleigh used on their Sports model came with an 18 tooth cog that gave a normal of 66.4 inches (for metric users, a ‘development’ of 5.3m) with a low of 49.8 (4.0m) and a high of 88.6 inches (7.1m). People buying a bike thought that three-speeds “didn’t have enough gears to climb hills and were too hard to pedal.” Raleigh Sports bicycles weighed 36 lbs and came with fenders and needed lower gears. The gears it came with were too high. I remember I used to routinely swap-out the cogs on the over-geared Raleigh Sports. If Raleigh had specified hubs with 22t cogs that would have given the ideal gearing with a medium of 54.3 inches (4.3m) with a low of 40.8 inches (3.3m) and a high of 72.5 inches (5.8m). As a result of being over-geared hub gear owners found themselves shifting often between normal and low with high gear being “way out there”. Probably more than everything else combined, this one detail made most customers prefer to buy ten-speed derailleur bicycles.
The only explanation I ever heard for why Raleigh over-geared their three-speeds was that cyclists used to turn higher gears with longer cranks at lower RPMs. Still, it never made sense.
Along about this time another seemingly unrelated trend was to unmask one of S-A’s characteristics as a defect: gear slip. Ten-speeds were built for a full-tuck riding position and were awkward to ride. The only accommodations made for women was to tilt the saddle nose down 2-3 degrees. To people with chronic back problems I repeated the findings by French physicians that the full-tuck riding position was actually good for your back. To those with carpal tunnel problems I recommended gloves. For men with prostate problems I recommended a leather saddle. Like the ordinary bicycle, or penny farthing it replaced, the derailleur bicycle had evolved to serve a narrow bandwidth of young, athletic, male customers. The bicycle was difficult to ride and maintain.
The emergence of mountain bikes changed all this. With an upright position anyone could easily ride them, and with an emphasis on durability they brushed the flimsy ten speeds aside. S-A gears could not be used for off-road riding for two reasons, the foremost of which was gear “slip”. If the hub is not in adjustment and you are riding in either normal or high, the gear can unexpectedly go into a “no gear” intermediate position between the gears. When it does this the gear suddenly disengages itself under load. Gear “slip” had long inhibited “honking”, or riders getting out of the saddle to pedal up hills because there is always the chance that while you are standing on the pedals it could suddenly slip out of gear. Since mountain biking required out-of-the-saddle pedaling for climbing hills S-A gears were not an option.
Another problem was gearing; S-A recommended that the rear sprocket be no larger than 22 tooth be used (presumably with a 46 tooth chainwheel) giving a 54 inch (4.3m) normal with a low of 41 inches (3.3m). Lower than this and, and as Frank Berto once said “presumably, the hub would grind itself into pieces the size of tooth fillings”. Mountain biking needed these lower gears, especially since honking was not possible. Think about it: you couldn’t use ultra-low gears but you also couldn’t stand up to pedal. This effectively shut S-A gears out of mountain biking, limiting them to road use.
Because the bikes they provided for road use were all hopelessly over-geared the picture of S-A’s inevitable obsolescence is more understandable. What is most curious is that S-A hubs had been “slip-free” since the 1904-1937 “X” model hub. Why didn’t they just bring back “slip-free” hubs when mountain biking became popular and why did they ever quit making such hubs in the first place? The answer has to do with the series of management companies that ran S-A.
From the beginning S-A was owned by Raleigh Cycle Company. Frank Bowden, Raleigh’s founder, was in the process of building what was to become the world’s largest bicycle manufacturer when he was approached by William Reilly, a poor Irishman, about a three-speed hub he had just invented. In 1902, in a series of legal maneuvers he swindled William Reilly out of the patent rights and got rid of him (Hadland, 1987, p.52). Reilly died in obscurity on a curb in Stockport, near Manchester in about 1950 when he was 83 (Hadland, 1987, addendum); S-A was thus born with Raleigh was the management company.
A Lack of Innovation
The original 1902 hub Reilly designed was a fixed-gear three-speed; you could shift gears but you couldn’t coast. It had an external freewheel threaded on, so you could coast. It had two “intermediate” no-gear positions between the gears to prevent simultaneous engagement of two “fixed” gears, wrecking the hub.
In 1904 the “X” hub replaced the original design. It didn’t need an external freewheel to coast because inside the hub it had three sets of pawls, one set for each speed. If simultaneous engagement of normal and low gears occurred the normal set of pawls was turned faster than the low gear pawls and over-ran them. If there was simultaneous engagement of normal and high the high gear pawls over-ran the slower normal gear pawls. A loose cable gave low, so if the cable broke the hub stayed in low. Designed by the inventor of three-speed hubs, William Reilly, BSA, continued to produce this hub for their bicycles until 1955 when they were acquired by Raleigh who discontinued Reilly’s hub. Bowden took over management of S-A in 1909 after he got rid of Reilly.
Bowden was a brilliant businessman in the process of building a bicycle manufacturing empire. He was not as interested in innovation as he was holding down costs. Raleigh allowed S-A to only just barely survive. Raleigh didn’t want anything but over-geared three-speed hubs and as long as that’s all they produced they didn’t care. The creative genius of S-A engineers was reduced to cutting costs to extend their tiny budget. I always imagined the R&D department as something like Hitler’s bunker.
WWI, production and design problems plagued S-A from l9l4-1918, when they designed the K model hub. This got to the market in 1921. S-A designed it as a cost-cutting measure so that if the cable broke the hub was stuck in high. This was a step backward. For another thing the K hub had only two sets of pawls. One set was used for both high and normal speeds and the other set for low. In high gear the high/normal pawls were fed after being multiplied by the planet gears for high gear. In normal same set was fed directly without being multiplied by the planet gears. In both gears the low-gear set was over-run. For low the sliding clutch “tripped out” the high/normal gear pawls unmasking the previously over-run low gear pawls. This method of “tripping-out” pawls prevented normal and low from being simultaneously driven. There was a danger, however, of simultaneously engaging normal and high. This was prevented in the following way. The fronts of the six clutch arms were square to fit against tabs on the inside of the gear ring for normal. The backside of the six arms of the sliding clutch were ramped so that if both normal and high were simultaneously engaged the gear ring was driven by the high gear. The drive of the faster turning gear ring drove itself into the ramps on the backs of clutch arms and pushed the clutch into full engagement with high. While clutches were expensive to make and tended to wear out there was no “no-gear” position and the hub was always in gear. In 1937 the K hub was replaced by the AW model with the infamous “no-gear” position (see Figures 2-7 below). AW clutches had four unramped arms and were cheaper to produce (Hadland, 1987, p.97). Like the K hub it replaced the AW used two sets of pawls and tripped out the normal/high set to unmask low, preventing simultaneous engagement of normal and low speeds. To prevent simultaneous engagement of high and normal AWs had a “no-gear” intermediate between them to keep them separate (Hadland, 1987, p.97).
It was not so much that S-A was unable to lead the field in market development as it was unwilling.
S-A continued to patent two additional different “no-slip” designs in 1948 (Hadland, 1987, p.120) and 1971 (Hadland, 1987, p.157). Raleigh patented their own “no-slip” design in 1972 (Hadland, 1987, p.154). Raleigh blocked production of all these.
Another example of S-A’s inability to lead in product development is hub brakes. Consider the popularity of disc brakes now. I recently opened a bicycle mail order catalogue and counted no less than six different kinds of disc brakes. When downhill racing became popular S-A announced they felt the disc brake “had no future in cycling” and stuck with drum brakes. In l985 when S-A said their gears were not strong enough for mountain biking (Hadland, 1987, p.168) it was reminiscent of the 192Os and their making similar disclaimers that their gears were not strong enough to be used on tandems. While S-A had produced a tandem three-speed in 1934 (Hadland, 1987, p.90) but they deleted it in 1941 (Hadland, 1987, p.189), giving derailleurs a niche market in which to get a toe-hold after WWII.
As the world changed the circle of Victorian Engineers at Raleigh who seemed to run S-A rigidly refused to acknowledge the world had changed since l9O4. They continued to make gentlemen’s gears for Edwardian cyclists. They probably figured that mountain bikes, like tandems, were just a fad.
Even in “gentlemen’s gears” they shunned innovation. Just like the “no-slip” three speed, S-A continued to patent innovations that Raleigh withheld from the public. Henry Sturmey patented a five-speed in 1921 and S-A continued to patent different designs of 5-speeds in 1940 (Hadland, 1987 p.111), 1973 (Hadland, p.155) and a 6-speed in 1954 (Hadland, p.130). Production of 5-speeds did not begin until 1966 (Hadland, 1987, p.146.); a 45 year interval.
S-A was starved and their resources plundered while Raleigh continued to grow. With over 7,000 employees in 1960 Raleigh was purchased by Tube Investments (TI), a manufacturing conglomerate that made everything from industrial tubing to several highly successful kitchen pans (Hadland, 2000). S-A had a new management company to run it and it was Raleigh’s turn to be managed. TI continued Raleigh’s regressive practices of plundering profits and blocking progress but had their own cruel twist about the mouth. Under TI’s management, Raleigh didn’t bring out a children’s hi-rise, or Stingray, bicycle until the demand in America was over. During the 1960s, Alex Moulton designed a small wheel bicycle and offered to sell it to Raleigh. After rejecting his offer Alex Moulton made them himself (Hadland, 2000). The bicycles proved wildly popular and Moulton bicycles quickly became Raleigh’s #1 competitor (Hadland, 2000). Raleigh delayed bringing out a BMX bike until it was too late to develop market share and got such a late start in mountain bikes (1984) they lost millions (Hadland, 2000). Now it was Raleigh’s turn to have its hands tied.
One outstanding example of the way TI prevented innovation involved the geared Dynohub. The Dynohub was a hub that contained a generator that powered bicycle lights. S-A had patented a geared Dynohub in 1967 (Hadland, 1987, p.147). While the generator turned as fast as the wheel a “geared” Dynohub generator was designed to turn faster than the wheel and generate more power. The Dynohub had been in continuous production since 1945 (Hadland, 1987, p.167). TI blocked bringing out the geared Dynohub. Rather than coming out with an updated more powerful model to stimulate sagging sales, it was cheaper to discontinue it and the Dynohub was withdrawn in 1984 as well as its battery-takeover option (Hadland, 1987, p.133). This was too bad because generator hubs are currently made by three different companies; Schmidt, DT and Shimano.
Another example of cost saving measures was the indicator chain coming out of a hollow axle, a feature of all S-A hubs. Because you couldn’t use a quick-release it had all the disadvantages of both a quick-release hollow axle and a solid axle; without the strength of a solid axle it was weak like a hollow axle but without the convenience of a quick-release. S-A patented a solid axled 7 speed in 1974 (Hadland, 1987, p.157). The 1974 patent on the solid axled 7 speed had expired by the time Shimano began producing S-A’s designs and producing gears with solid axles. By the time S-A closed Shimano had a solid axled 4 and 7 speed, Sachs had a 5, 7 and 12 hubs that used indicator chains and Rohloff had a 14 speed in both solid axle and quick-release versions, before S-A finally came out with the solid axled 7 speed they had patented 20 years before.
As early as 1984, as Mountain bike sales exploded, S-A, under TI, produced a “no-slip” three-speed for Columbia Bicycles in America that was not released to the general public. It had three sets of pawls and a ramped clutch. In 1984 Raleigh still didn’t have a mountain bike. By Feb. 1987 they had made 60,000 hubs for Columbia (Hadland, 1987, p.167). They were still in production in 1997 though they had not been released in the United Kingdom (Read, p.114). Their steadfast refusal to issue these to the general public is but one of many decisions that led to Sturmey-Archer’s increasing obscurity.
Moreover, three speeds, for all their faults, actually cost slightly more in America than their lighter ten-speed counterparts in the late 70s. For example, the Raleigh “Record Ace” introductory ten-speed was $265 in 1979 in the USA. By comparison a 3-speed Raleigh Sports cost $285. While the emerging mountain bikes still cost well over $600 prices were shortly to begin dropping rapidly as they exploded in growth.
Though British-made Raleigh imports to the U.S.A. stopped in 1981 there were a lot of three-speed bicycles still around and I could still find work as a Sturmey-Archer specialist.
About this time Sturmey-Archer was forced to discontinue rather than update a number of products which had long been in production such as the Dynohub with its battery take-over. They redesigned the drum brake which they offered in a bewildering array of materials and finishes.
The road to obscurity had not only to do with mountain bikes and S-A’s withholding from the public the same improved designs they provided to manufacturers but also a history of defective designs for products they did release to the public that betrayed the loyalty of even the most die-hard customers.
In 1980 I convinced a shop to order a couple of 5-speed hubs. Someone heard we had them and drove fifty miles to buy one. But he kept bringing it back saying it “slipped” out of gear. We ended up giving him his money back. A few months later we got a notice from S-A that the hub had a “faulty” spring. It was too late because no bike shop in town would take the chance on selling five-speed hubs. Some time later I found the shift levers it came with were also “faulty”.
I remember being delighted when S-A came out with an aluminum-alloy-hubshell five-speed hub in 1983. This was withdrawn in 1989 when it was found that pieces of the gears would burst through the hub shell (Read, p.83). Though I never saw this it must have been spectacular to see. The 5-StAr hub (the capitalized “A” playing on the first two letters of the words “Sturmey” and “Archer”) was introduced in 1991 but was withdrawn two years later because it tended to break axles. Though these happened on a relatively minor scale they had the effect of alienating the finicky and touchy market that cyclists are. S-A came out with a series of triggers that, with their increasing reliance on injection-molded plastic, bore an uncanny resemblance to a child’s play-toy. No serious cyclist would even consider putting such ugly junk on an expensive Reynolds 531 frame.
The real tragedy of this is that S-A didn’t have to fail. Recent research by Frank Berto and Chet Kyle indicates that run-in and well oiled Sturmey-Archer hubs are 91.8%-95.6% efficient compared to a Shimano derailleur’s 86.9%-95.9% efficiency (Berto & Kyle, 2001).
In 1979 S-A completely dominated the market. The history of bad design choices has its roots in some of their earliest designs. Before 1922 if a cable broke on a S-A hub it was locked in low (Hadland, 1987, p.74). Under Frank Bowden they redesigned it in 1922 so if the cable broke the hub was stuck in high. They brought out the “no gear” position in 1937 because it was cheaper to make. In 1954 they substituted the AW three-speed design with the ill-fated “SW” model three-speed (Hadland, 1987, p.134) with springless crescent-shaped pawls. How this hub got into production is curious because it never worked and slipped in every gear. What is most surprising is that it took them three years to withdraw it! I don’t know if they replaced all the defective hubs they sold but I don’t think they did.
At the time I first started working in a bike shop Sturmey-Archer was owned by Tube Investments which also owned Reynolds, Raleigh and Brooks. Like a ship breaking up on its way to the bottom Tube Investments sold Raleigh USA to Huffy; the largest manufacturer of junk bicycles in America in 1982. Anticipating a corporate sell-off, Derby, an American management firm was formed by a wealthy American tax attorney expressly for the purpose of acquiring old Raleigh holdings as they were sold by TI (Hadland, 2000). Raleigh of England, Reynolds, Sturmey-Archer and Brooks came under the management of Derby, Inc. in 1987 (Hadland, 2000).
TI had pretty much ruined Raleigh/S-A. By the time TI was finished Raleigh went from employing 7,000 employees in 1960 to 700 in 1987 (Hadland, 2000). The factory was virtually the same as when they bought it, only more decrepit. Moreover, aluminum was the tubing of preference and Raleigh couldn’t do anything but make steel bicycles.
S-A had their resources plundered and development choked for 85 years by first Raleigh and then TI. Derby, their new owner, would eventually own Raleigh of England, Raleigh of America, Nishiki, Kalkhoff, Univega, Gazelle, CyclePro, Haro and Diamondback and be the largest manufacturer of bicycles in the world. Under Derby, S-A began innovating and was allowed to finally release their “no slip” three speed hubs, the ill-fated 5-StAr in 1991 and the solid-axled 7-speed hub to an optimistic public in 1995 (Read, 2003). Tragically, both these designs were defective and S-A had to issue replacements. In 1999, after twelve years of trying to undo the damage and upgrade the factory Derby quit trying. They sold the land Raleigh and S/A were on to the University of Nottingham and auctioned off Raleigh’s brand new laser cutting equipment and robotic machines. They didn’t know what to do with S/A’s buildings and patents and so they sold S-A and Brooks for the price of a three-speed hub (Hadland, 2000). The buyer was Lenark, a shady British management firm who had previously been investigated for fraud.
The law of supply-and-demand was never so cruel.
The end came abruptly. Shortly before my birthday, in October, 2000, the employees were called in, told that Sturmey-Archer was closed and they had ten minutes to get out. A year later I still had trouble accepting that Sturmey-Archer was closed, the workforce unemployed, the land sold to a university to train people for a future without jobs, the buildings knocked down and the machines put in a ship container for Taiwan where Sun Race, their newest management company was located. For me, S-A is gone forever.
S-A’s failure had its origins in 1904. What we are seeing is the dark fruit of greed. The greed that designed hubs for cyclists like the SW and the AW with its “no-gear” position was the same greed that sold the land S-A was on to a University. Greed and avarice do not serve society any better than they serve the needs of cyclists.
My own fortunes followed those of Sturmey-Archer’s. While I continued as a mechanic I was not hired as a Sturmey-Archer specialist after 1985. The bike shop I work in now is a Raleigh shop. It has one extra three-speed cable and one small plastic drawer labeled “Sturmey-Archer” containing a couple of frame fittings.
Berto, F. and Kyle, C., (2001), HUMAN POWER, #52. Summer, 2001. pp.#3-11. Though other articles have appeared measuring the efficiency of epicyclic gearing this is the most recent and possibly most carefully done.
Berto, F., Shepherd, R., and Henry, R. (2000), THE DANCING CHAIN, Self-pub., San Francisco. The definitive text on derailleur design and history. Imprimatur.
Hadland, T. (1987), THE STURMEY-ARCHER STORY, Self pub., s.l. The definitive text on Sturmey-Archer hub gear design and history. Nihil Obstat. Imprimatur.
Hadland, T. (2000), Raleigh in the last quarter of the 20th century. In the 11th INTERNATIONAL CYCLE HISTORY CONFERENCE PROCEEDINGS, A. Ritchie & R. van der Plas, Eds. Van der Plas publications, San Francisco, Calif., USA. Chronicles the confusing business dealings that led to Raleigh’s leaving England and Sturmey-Archer’s closing.
van der Plas, R. (1991), BICYCLE TECHNOLOGY. Bicycle Books. San Francisco.
Read, P. M., STURMEY-ARCHER: “THE END” AND “NEW BEGINNINGS”; (1902-2000 onwards) hub gear drawings, diagrams and parts lists, Fourth Ed., (1997), Self pub., Milton-Keynes. Peter Read has a business repairing Sturmey-Archer equipment. This massive tome is the definitive guide to repair each S-A design with many small changes carefully documented. It includes the many aftermarket improvements that Sturmey-Archer cognoscenti have developed over the years.
The author thanks Tony Hadland on whose research this article is almost entirely based and for his invaluable suggestions. The author also thanks David Gordon Wilson and Peter Read of Phoenix Hub Gear Repairs for their invaluable comments on earlier versions of this paper. The author is especially grateful to Marv Wells, in Columbia, Missouri for his expertise in preparing the images. You’re a good bud, Marv.
The author is a bicycle mechanic in Columbia, Missouri, U.S.A. You can contact him by email at firstname.lastname@example.org or by paper mail at 1007 Grand Ave., Columbia, MO 65203-4025, USA or by telephone on + 573-442-1187
Figures and Text both copyright (c) by Vernon Forbes, 2003. (Minor revisions February 2012)