BEARINGS AND POINTS OF CONTACT.
Friction gives us a grip on the earth, and is indispensable for
propulsion, but it is not in the least wanted in cycle bearings or in
any other bearings, and one of the problems of mechanics is how to
reduce it as much as possible in places where it consumes power as well
as produces wear.
No material thing, however polished, is quite smooth; every surface
may therefore be considered as covered with irregular hooks or teeth,
however flat and smooth it appears to the unassisted eye, and these
catch and hold one another, producing the hang or drag called friction.
Oil, being a fluid, fills up the spaces between these invisible teeth
and levels off the surface; the office of lubrication is, therefore, to
get between the contact surfaces and keep the hooks or teeth thereon
from touching. When surfaces are desired to slip and slide on each
other easily, oil is helpful; when the hooks or teeth are to catch into
each other, as between locomotive driving wheels and the rail, grease
is out of place, whether it is oil or grasshoppers, for it spoils the
“adhesion.”
The earliest mode of reducing friction is doubtless as old as the
Tower of Babel, for the idea must have occurred to the primitive
man. It is simply to put a roller underneath and convert sliding
into rolling motion. This is in principle equivalent to mounting the
weight on wheels, and it is the solitary and final way of dealing with
the problem of friction. The common grindstone bearing is a familiar
example; the axle of the stone rests on the rims of a pair of small
wheels which stand so as to lap past each other. Here the axle rolls
the wheels as it turns, and their motion at their centres is so slight
that friction is nearly eliminated.
EVOLUTION OF THE BEARING.
In cycle construction, the first bearing was the “plain” one in common
use elsewhere; then a nicely fitted and hardened sleeve was added, and
this was known as the “parallel” bearing. Rollers were also tried, but
rollers have a determined habit of going askew, one end moving faster
than the other, and as soon as they get out of parallel thus, they set
up a great resistance. To meet this difficulty, the ends were sometimes
made so as to overlap and match into one another, or the ends were
loosely passed through thin rings, which revolved with the rollers
around the axle; but the rollers still tried to run askew, and the
efforts of the rings to prevent them caused another friction, so that
the roller was abandoned. About the same time, the “adjustable cone”
was tried. This was a male cone, threaded on the axle and fitting into
a female coned space in the hub. The character of the rubbing action
was not changed by this device, which was called a device to make wear
in order that wear might be taken up, but the parts could obviously be
kept in contact (though not in nice fit) by screwing the cone further
in.
The next and final step was to interpose steel balls between these
coned faces; and as the ball is a very short roller, with ends rounded
off, it can go in any direction it pleases. The principle of lateral
adjustment by moving a coned surface to or from another coned surface
opposed thereto, with balls placed between, was patented more than
twenty years ago and is still in universal use; yet, as just remarked,
this is the adjustable cone modified. It is to be borne in mind that
the only possible service of the cone, as before, is still to take up
wear, and also that the retention of the cone for adjustment introduces
new difficulties. Note also that on the old high “Ordinary” the large
wheel had its bearing cases fixed and the axle revolving, because the
power was applied to the axle, while the rear wheel had its axle fast
and the wheel hubs revolved around it. On the modern bicycle the method
reverses, both wheels revolving around fixed axles, while the crank
shaft, which is the part receiving the driving power, revolves within a
fixed bearing-case.
The revolving axle used to have two grooves, matching grooves within
the fixed case, and the balls were held in holes in loosely fitting
rings which slowly travelled around with them, these rings having no
use except to aid while putting parts together. This double row bearing
was called non-adjustable, because the sole way of tightening it up
was to move the two halves of the bearing-case closer together; for
this purpose the case was made in halves, as a “split lug,” and held by
screw bolts. Yet this construction, if well made, solved the difficulty
of the “points” in bearings and gave the balls a correct rolling motion.
THE QUESTION OF “POINTS” OF CONTACT
The ball may be regarded as a number of tiny thin wheels or disks,
borne on a common axis. Obviously, the larger the wheel the more
easily it will roll; hence we reach the first rule, namely: the
ball should rest and roll on its largest diameter, if possible, and,
as a corollary, large balls (within reasonable limits) are better
than small ones. In order to fully carry out this rule and use the
largest diameter, the ball must be placed between two plain cylinders
or rings, and the weight must bear in a direction at right angles to
the plain surfaces; the ball will then roll at its best, and yet this
construction is not practicable. This is so because there would be no
means of keeping the balls in one track and because the surfaces and
the balls would not stay in contact, there being no “adjustability”
or means of moving them closer together. Coming, then, to the usual
construction of a fixed axle having on it a stationary cone, and a
wheel hub revolving about this, we reach the important practical but
not half-considered question of “points.” That is, on how many points
in its surface shall the ball rest? The hub is commonly called the
“case” or the “cup.” If the ball rests on the cup at one spot and on
the cone at another, the bearing is called “two-point,” or “spot”
is more nearly accurate than “point,” if by the latter the literal
mathematical point is meant; if the ball rests on the cup at two places
and on the cone at one, the bearing is called “three point;” if the
ball rests at two places on cup and cone both, the bearing is called
“four point.”
[Illustration: OLD DOUBLE-ROW BALL BEARING.]
Referring to the cut of the two-point, it is plain that one of the
coned surfaces shown, revolving in a plane at right angles with the
axle, must roll the ball on the other cone, the ball running on both
in planes parallel to the plane of motion of the revolving cone, as is
indicated by the dotted lines; hence the ball will roll, and not slip
or slide. To a very limited extent the two-point bearing has been used
in this country. We can at the moment name only one make which we are
sure has had this form really so made, and well made, with the surfaces
accurately curved so as to place the balls correctly and with grinding
after shaping. This make is the Humber, which deserves honorable
mention for the importance attached to the bearings and for the
intelligent care with which they have been constructed. This remark,
however, is by no means meant as exclusive or as implying that no other
makes have excellent bearings.
[Illustration: LOWER HALF OF DOUBLE-ROW BALL BEARING.]
An interesting form of two-point bearing is the Lake, made by the C. S.
Caffrey Company of Camden, N. J. It makes the coned faces of cone and
cup parallel and flat, inclined at an angle of 45 degrees to the axle.
Here it is evident that the ball will run without twisting or skewing,
and in order to keep the balls in place the old device of putting them
in a perforated loose ring is employed. The holes in this ring for the
balls are made oval instead of round, in what does not seem a very
well grounded expectation of thus removing the slight friction between
ball and ring. The holes are also “staggered,” so that the balls do
not run on exactly the same tracks. It is claimed that, on a test, a
front wheel with this bearing, being whirled by the hand, ran an hour
and five minutes. This must be admitted to be a remarkable performance,
even if the adjustment were loose.
Far the commonest construction, however, has been the three-point,
partly because, by a confusion of ideas, a three-point bearing has
seemed as if it must be firmer than a two-point, and partly because
the former can be turned out at a very moderate cost. As in almost
universal use during several years past, and as produced by the
parts-makers almost without exception, the form of this is as shown
in the cut. (See page 86.) Turn the page so as to bring the surface C
on the cone horizontal, and if you then imagine this surface C in the
same plane as line CD, it is easy to see that the ball will roll upon
the case at A and B both; and as the diameters of the ball at A and B
are equal, it will roll around the circle easily and without skewing.
As the inter-action of the parts is not changed thereby, we for the
moment, as a matter of convenience, assume that the cup is stationary
and the axle turns, which is the reverse of the fact. In actual
position and working it is evident that under the weight of the load
the ball will slip down the slope at C and be pressed hard against the
side B as well as against the bottom A. The relative pressure on these
two points will depend on the flatness or steepness of the surface
C, but ordinarily the pressure on the two will be nearly equal. The
action at C tries to roll the ball on a horizontal axis, parallel with
the wheel axle; the action of B upon the ball tries to roll it on a
vertical axis, parallel with CC. Moved by C, the ball may roll on A and
slide on B, or it may stick fast to C and slide on A and B both, or it
may stick fast to both A and B and slide on C. Certainly it cannot have
more than one of these movements at any time, and hence the ball cannot
possibly roll in two directions at once.
To make this more clear, imagine the ball and the two surfaces to be
toothed where they come in contact, thus being visibly gear wheels; if
these teeth are spur-teeth, the cone will impel the ball in its own
plane of motion, namely, line CC, and the ball will then roll on side A
and rub on side B; if the teeth are bevel, the ball will roll on B and
rub on A.
[Illustration: “FOUR-POINT” BEARING—TWO POSSIBLE FORMS.]
[Illustration: ACTUAL “FOUR-POINT” BEARING—“E AND D” PATTERN.]
HEEDLESS CONSTRUCTION.
For this reason—that this “jammed in a corner” pattern of bearing
requires the ball to perform a physical impossibility—it must be
unsparingly condemned. Indeed, if there is one form of polite
and parliamentary phrase more decisive than another, we wish to
be understood as using such form in condemning this particular
construction. It does not violate any statute law, but it does violate
laws of mechanics and good sense. What the ball actually does under
such conditions is to “get around” as best it can, rolling somewhat,
sliding somewhat, and slipping and skewing between times. The balls
rub a little on each other and their contacting surfaces are moving
in opposite directions; hence it is not to be supposed that they
invariably roll, under even the best conditions, the only certainty
being that they always follow the line of least resistance. Here we
might say that exhibitions of a transparent bearing on a large scale,
such as were at the recent shows, amuse visitors but prove little, and
yet a close scrutiny of them will show that the balls have an irregular
action; moreover, such a device as the “dynagraph,” professing to show
graphically on an indicator card the frictional resistance of bearings,
is a waste of ingenuity and construction, because it cannot be worked
under actual practical conditions as when the wheel is in use. The
difficulty with bearings as generally made hitherto has been that
notwithstanding much talk in catalogues about “tool steel” and smooth
grinding the common way has been to press the cups into the hubs, screw
cones on the axle, drop in balls, turn up to place, and let it go so.
Even in 1898, many catalogues furnish no information, either by text
or by cuts, as to construction of bearings, and when we have had no
other means of knowledge it has been in not a few cases impossible to
find out certainly even such a distinct and practical matter as whether
the adjustment is “cup” or “cone,” in such a heedless way has this
part of the bicycle been passed over. Makers have been too prone to
count anything with balls and a cone as a ball bearing, and they have
had a good degree of liberty allowed them to so consider by these two
facts: the rider does not know and the repairman does not care, and if
a bearing is not screwed up too hard and run entirely dry it will move
with a fair degree of ease even though the balls can not roll much. And
yet in all such cases the defect makes its own witness by the “flats”
made on cone and balls and by the ball track cut into the cup.
BALL-MAKING.
About eighteen years ago Col. Pope said to the writer, referring to the
first Columbia, then in market and the first American product, that
it would cost $25 to put ball bearings on the back wheel (or possibly
it was on both wheels). The usual extra on English makes at that time
for balls to back-wheel was one pound sterling; the first ball pedals
were also expensive, but for some years past any bearing without
balls, even on the lowest-priced wheels, would have been rejected by
every buyer. The difference has come largely by cheapened processes of
ball-making, and, as in other things, reduction in cost and betterment
in quality have come together. There are several ways of producing
balls. According to one of the best, the Simonds Rolling Machine Co. of
Fitchburg use forging machines, which are substantially two uprights,
a half-die on each upright, and work automatically. Heated rods of
tool steel are inserted in this machine, which forges a ball rough
and at the same moment bites off the bit from the rod with the die.
Next follow grinding and polishing automatically between horizontal
disks about three feet in diameter in conjunction with emery wheels;
finally come tempering, the last polishing and gauging automatically.
Ordinarily a maximum variation of 1/1000 of an inch has been considered
close enough, but this Company are able to guarantee a variation not
over 4/10,000, the highest accuracy and uniformity being naturally
considered somewhat in the price. The machines used are patented, and
this bare outline is all we are permitted to publish.
There remains to be considered the four-point bearing, and no better
example of this can be given than in the cut of one as used on the used
on the “E. & D.” as made by the Canadian Typograph Company of Windsor,
Ontario. It is proper to say here that only minor details on this are
patentable, for the principle is old and was in the old Bown Eolus
bearing as long ago as 1877. Reference to the cut shows clearly that
the ball rests on two points on cone and cup each, that its diameters
are equal at these places of contact, and (most important of all) that
the direction of pressure on the ball is at right angles to the axle,
and hence that the ball will roll on an axis parallel to the axle;
therefore there can be no sliding or skewing.
The contact surfaces are a right angle V in section, or can be made by
cutting open a square diagonally. At the last show in New York, as a
test, ten single wheels of this make were suspended in pyramidal form,
and these were all run, day after day, by a single length of No. 100
sewing silk. A wheel was also shown with the balls removed from one
bearing and tightly screwed in the other. This wheel was then whirled,
being supported by one end of the axle on the finger, gyroscope
fashion, to show the extraordinary absence of friction.
THE “CUP” OR “DISK” ADJUSTMENT.
Of the highest importance is also “cup” adjustment, as opposed to the
usual “cone.” As seen in the cut of the three-point bearing, in the
latter form the cup is pressed into the hub and stays fast there,
as a seat for the balls, with its coned surface facing outward. The
adjusting cone faces inward and screws on the axle. The “cup” form
reverses this, facing the axle cone outward and leaving it fast on the
axle; the cup faces inward and adjusts by screwing into the hub end, as
shown in cut of the bearing last described. The practical advantages
of this method are very real and are these, as may be indicated by the
somewhat rude cut:
1. When the cup is pressed into the hub it may not be quite
true across the axle line—all the more if it is afterward
removed and replaced, as in changing the rear sprocket. Any
nut has some degree of side-to-side movement on its thread,
however nicely fitted; an adjustment cone on the axle can
also never be held quite firmly in position, and the grip
of the fork ends upon it is even liable to cant it to one
side to the slight extent of its looseness in the thread.
This interferes with accuracy in the bearing. But the cup is
of larger diameter and hence is steadier in the fit of its
thread, and it is also practicable to lock the adjustment
more firmly on the cup. Moreover, in the “cone” form the
cone has to run on the same thread with the outer nuts which
hold the wheel in the fork, and this thread must be coarse,
because a fine one would not have sufficient strength to bear
the strain of locking the wheel; so the adjustment must be
on the coarse thread. But when the cup screws into the hub
as proposed, it has its own separate thread, which may be as
fine as desired; thus it gains in steadiness by fineness of
thread as well as by larger diameter.
2. Dirt cannot enter through the thread, but only through
the open joint. The opening close to the axle is obviously a
smaller circle than at the edge of the movable cone on the
old method; hence the cup form excludes dirt better, and if a
felt washer is used the friction from that is less when put
close to the axle than when farther away.
3. The cup adjustment has the great and obvious advantage
that the adjustment is wholly independent of the fork, being
only on the hub itself; hence the adjustment can be made
more easily and accurately, and after being so made once for
all the wheel can be removed and replaced without danger of
disturbing it.
4. In the other form of hub, oil naturally runs out; with
the cup adjustment, the parts are readily and naturally
arranged so that oil is held at the bottom as in a reservoir,
and the balls can run in it. Reservoir hubs of this pattern
are quite well known in England, and the makers of the “E.
& D.,” who use a felt washer as indicated in the cut, claim
such a perfect exclusion of dirt and retention of oil that
the latter is found still in its place at the bottom, not
discolored, after over a thousand miles’ running.
In England the cup adjustment has long been standard, although perhaps
not invariable; the tardiness of its adoption in this country must be
ascribed to an insufficient study and appreciation of the practical
importance of bearings, and to the considerable investment already
made in parts and tools on the other plan. The first step in adopting
the cup form here was at the crank-hanger, where it has been quite
largely in use for several years; but it is being applied to the
wheels as well, and among the makers using it all over we note such
well-known concerns as the Liberty, the Sterling, Humber, Victor,
Howard and Lyndhurst. This is not an exhaustive list, for we have not
studied every catalogue; moreover, it is impossible to determine the
point in all instances, and many who do not use this form on wheels
have it on the crank bearings, as also some others screw the cup
into the wheel hub and “back out” the axle cone for adjustment—this
last is good as far as it goes and is a half-way step. After having
constantly advocated the cup adjustment for several years past it is a
gratification to find it thus making progress, and we note this as the
chief step in improvement of bearings in 1898.
GENERAL IMPROVEMENT IN BEARINGS.
Yet it should be said that there is betterment in bearings generally—in
accuracy and temper of balls, in fitting and grinding of cones and
cups, and also in the means of adjustment. But excellence in details
may also have some effect to conceal errors in plan, and it should be
clearly noted that easy spinning of a bearing may even mislead. The
parts being hard and smooth, and oil being present also, the balls
will get around with slight resistance, whether rolling or sliding;
but the test comes only under load, especially under the heavy strains
which tend to cross-twist frames. The two-point bearing, provided it
is really designed and made in the best manner as such—and the proviso
means a good deal—will work satisfactorily; the three-point also can
be so designed and made as to allow rolling of the balls, although it
is less facile and manageable than the others; the four-point is the
best theoretically and seems easiest to construct. The “corner” pattern
we have felt obliged to condemn will “go” after a fashion, as above
admitted; but bicycle evolution is toward uniformity and simplicity,
and as it has been proved just as economical to construct right as to
construct wrong, after the preparations are once made, there ought to
be positive insistence on one thing always, and that thing “the best.”
[Illustration:
“THREE-POINT” BEARING: COMMON FORM AT LEFT—POSSIBLE
IMPROVEMENT AT RIGHT.]
LUBRICATION AND DUST EXCLUSION.
It is always a mistake to suppose that even a ball bearing can
successfully and wisely be run without lubrication, under load,
although the feasibility of so doing has been declared on what ought
to be pretty fair authority. But the rolling movement can never be
made absolutely constant and the sliding perfectly gotten rid of; even
if this could be, the contact sides of the balls move in opposite
directions, and, therefore, must rub slightly on one another; if the
balls are held apart by a perforated collar they cannot touch each
other, but they touch the collar instead and rub on that—so a little
friction will remain in the most favorable circumstances. By the way,
some stick vaseline in bearings, but we must disapprove this; we do not
regard that substance as a lubricant in any proper sense, and the very
quality of adhesiveness which makes it convenient sometimes for holding
balls in place while assembling a bearing also makes its presence
objectionable after the assembling is done. It has the property of
staying in place and not flowing out, not being fluid; but this
property is possessed by other substances—by tar, for instance.
Of course, there must always be an open joint in every bearing (the
outer end of the pedal excepted) where a moving part passes a fixed
one. Nicety of fit, so that these two parts shall be almost in contact,
is the first requisite, and is not to be found on “cheap” wheels,
albeit such wheels (on paper) have tool steel and dust-proof bearings,
like all others. Hard-rubber washers cannot close these joints; soft
rubber would quickly be destroyed by the oil. Felt and velvet have been
the only recourse, and this not an entirely satisfactory one, although
if dust and mud could be perfectly excluded the oil might retain color
almost indefinitely. For this, at the risk of repetition, we must say
that the cup adjustment—called disk adjustment generally in England,
and sometimes here—is exactly adapted, and that the four-point serves
best. These soft packings are still retained by quite a number of
makers. For instance, the Monarch puts on the left side of the rear
wheel an octagonal-faced dust cap, and next to that a felt washer,
and next to that the usual ball retainer; the front hub has this
construction on both sides, and the crank bearings are fitted with cup
adjustment.
SOME DISTINCTIVE 1898 FEATURES.
The Crescent has a new mode of adjusting the bearings at the crank
bracket. A loose collar, with a projection which fits in a slot in the
edge of the adjusting bush and has its own inner edge finely scalloped
goes over the bracket; the lock-nut having been loosened, this bush is
free to turn to the right point, when the sliding collar is simply slid
back until one of its scallops catches on a pin set on the bracket,
and then turning up the lock-nut makes all fast. On the wheel axles, a
separate nut is added to lock the adjustment cones, so that this can
be done with the wheel either in or out of the frame, and tightening
up the fork nuts cannot affect the adjustments. Felt washers and ball
retainers are used throughout.
The Magnet Bicycle Company of Chicago, makers of materials and
fittings, offer in the Magnet hub a novelty in a combination of a
concave and convex lock-nut and washers, which, when the parts are
screwed home, avoids one of the defects of a cone adjustment, namely,
the liability of the axle cone to tilt or tip on its thread under a not
parallel approach of the fork end: the adjustment is also not affected
by handling the hub.
[Illustration: COMMON THREE-POINT BEARING.]
The Shirk shows several peculiarities. Not only are all bearings of
the cup adjustment pattern, but the axle cones (which are on a sleeve)
are two-faced and reversible, so that each hub has two reserve cones
or cone faces; moreover, as the axle is independent and serves only as
a support it can be pulled entirely through and out, after removing the
outside nuts, so that the front fork does not have to be spread to put
in or remove the wheel, and the rear wheel can be dropped out of the
frame without disturbing the chain.
It is quite the way to make the crank-hanger only a shell or a support
for the working parts within, without having them directly fastened
to it. The Phœnix follows this fashion by inclosing the bearings in a
separate sleeve, splitting the bracket itself on the under side, and
providing it with projecting lugs and screw bolts, so that it can be
pinched up to grip the bearing shell and hold it in position. But the
peculiarity goes further, for the opening underneath is so wide that by
turning these bolts out of the way and bringing the left crank opposite
the opening the cranks and axle with sprocket on—indeed, the entire
contents—can be slid to the right clear out of the hanger.
The Relay has a dust cap at the crank axle bearing, with a portion of
it consisting of translucent celluloid, and claims the makers’ catch
phrase “you see the balls.”
[Illustration: BALL CONTACT IN “TWO-POINT” BEARING.]
Besides using the cup adjustment, as already noted, the Howard—made
by the E. Howard Watch and Clock Company—has a peculiarity in that
the adjusting cone slides on the axle without being threaded. On the
crank axle is a nut working on a thread at the axle centre and bearing
against the end of the short sliding sleeve which constitutes the
acting cone. A set screw in this central nut is loosened by inserting a
wire or a nail through a hole in the bracket; the nut is held fast by
putting this wire or nail into a slot therein; then a slight turn of
the crank forward or back tightens or loosens the bearings at both ends
simultaneously by causing the nut, thus held from turning, to move the
axle to right or left instead; then the set screw is again fastened. A
similar nut is on the wheel hubs, and the wheel is turned back or forth
a little to adjust the bearings, the single nut at the left side of the
frame then locking the adjustment. On behalf of this peculiar device it
is claimed, with evident justice, that the operation is both quick and
sure, and that as the coned sleeve slides on the axle instead of being
threaded the bearings are bound to be true and in line. We ought to add
that although we have classed this form as a cup adjustment, it is not
literally quite so, the sole difference being that it belongs in the
class of bearings which face cones and cups in the way proper to that
form, but screw the cup into place once for all and adjust by “backing
out” the cone. Last year the Humber wheel bearings were of this type,
and so are the Lyndhurst now, although that make we have also classed
with the cup adjustment type, the difference being so small.
The Lyndhurst makers, by the way, while using the cup adjustment
strictly at the crank bracket, with an admirably made sleeve having
the cones slipped thereon, lay great stress—as relating to accuracy of
fit—upon making bearing parts “from the centre” in the old-fashioned
screw lathe, as against the monitor or automatic lathe; they aver that
the special hardness of tool steel forbids working it on the automatic
lathe, and that “there are not over six makes of bicycles in the United
States with centred axles, cones and shells turned from tool steel.”
Especial significance here attaches to the word “centred;” and every
cone is separately ground in a lathe upon its own axle.
The makers of the National of Bay City, Mich., have all bearings on
their best models, removable by sliding out intact, using also a
peculiar form of cup adjustment.
At the 1897 cycle shows, the Indiana Bicycle Company, makers of the
Waverley bicycles, exhibited their bicycles with cones sliding on
the axles instead of threading and screwing the cones or cups for
adjustment in the usual way. They used this system during the past year
upon many thousand Waverley bicycles, and the results have demonstrated
that this method is one by which absolutely true bearings can be
obtained; the cones and cups remaining always in the precise relative
positions in which they were ground, the cones in adjustment sliding
to or from the cups. In this season’s construction they have made a
slight change, however, and which may be regarded as a good step toward
the long sought for interchangeability in construction. The change
consists in having universal cones fitting either side of the front or
rear wheel hubs and having a double face with two ball races they can
be put on the axles in eight different places or ways, thus making it
impossible to assemble the bearings incorrectly. The construction is
also such that the bearings are as nearly dust proof as it is possible
to make them without binding friction. A bicycle adjusting cone or cup
that is threaded, no matter whether the thread be fine or coarse, must
necessarily have some sort of a spiral twist to it which prevents the
cone or cup from setting perfectly true as ground. The sliding method,
however, obviates this difficulty.
It is worth noting—especially as being a step toward simplicity and
uniformity in construction—that the Humber is now made with the bearing
cups of the rear wheel interchangeable with those of the crank hanger.
Another novelty in its way is the insertion of ball retainers in pedals
made by the American Watch Tool Co. The Sartus pedal, made by the
Warwick & Stockton Co., also uses a somewhat peculiar retainer.