SOME SPECULATIONS

SOME SPECULATIONS

NkT FAST-CAR SUSPENSION

—with particular reference to tha

t used on the German G.P. Cars

by SILdr. C. W. S. MARRIS, This masterful survey of modern suspension as to future trends and another article which technical matter of the "

systems represents both an interesting prophecy satisfy those readers who have asked for more . " sort in their " Motor Sport."—Ed.

IHAVE used the word" Speculations " in my title deliberately, because many of the categorical statements put forward represent merely my own amateur theories and may be scornfully rejected by the experts.

From late Edwardian days until the early thirties the suspension of sports and racing cars remained substantially unchanged, consisting of stiff felliptic springs on front and rear axles, controlled by friction shock-absorbers, in. combination with a fairly flexible channel steel frame. At the rear this was sometimes varied by 1-elliptic or cantilever springs.

The Lancia "Lambda," though not primarily a sports car, deserves mention as an honourable exception, its front end resembling in many essentials that of a modern G.P. car.

Springs throughout this period were extremely hard, that is, they had high rates and small excursions, and, being backed by large friction shock-absorbers, were extremely unyielding. Road irregularities were to a certain extent accommodated by deflection of the frame, and Bugatti even went so far in the early twenties as to build two or three experimental cars with no front springs at all. The reasons for using such hard springs were several. At the front hard springs were necessary for two reasons : (1) To limit vertical travel of the axle. With conventional Felliptic springing the front axle rises and falls in an arc whose centre is the eye at the unshackled end of the spring. If the travel is large, considerable changes take place in the castor angles of the king-pins, to the detriment of directional stability. (2) To minimise roll. Roll is produced by' centrifugal

force acting through the c. of g. of the sprung portion of the car with a leverage determined by the vertical distance of that c. of g. above the roll axis, and is opposed by the springs with a leverage determined by the horizontal distance between them (where an axle is used).

With a deeply dropped front axle and flat springs the roll a is at the front was low, and the minimum permissible distance between the springs was limited by the steering lock. To resist roll adequately hard springs were therefore necessary.

At the rear they were necessary in order to maintain good contact bet WeVII wheels and road in spite of high unsprung weights. The rapidity with which a spring rebounds into its previous position after an impact depends upon its rate, and upon the mass it has to accelerate, that is, the unsprung weight of its particular wheel, etc. The higher the rate and the smaller the unsprung weight, the swifter the rebound, and the sooner the wheel returns to contact with the road. Where the unsprung weight is large, as with the conventional live rear axle, the spring must have a very high rate if the wheel is to rebound swiftly enough to follow faithfully a rapid succession of small bumps. Where the unsprung weight is low a soft spring is perfectly able to maintain good road contact.

The use of stiff springs also relieved designers of the necessity of making special provision for brake reaction at the front and drive and brake reactions at the rear. Many fast cars up to the middle thirties were built without radius rods at the front and with Hotchkiss drive at the rear.

Such stiff springs are midesira hie. They give a harsh ride, and iMpose heavy Stresses on the frame 1)ecause the stiffer the spring the greater I he rate at which it transmits to the frame energy imparted to its road wheel by an impact. Also, the hard spring, with its small range of movement, c:innot accommodate large slow excursions of its wheel, such as are demanded by considerable but relatively gradual changes in level of the road surface. In 1934, Mercedes-Benz and AutoUnion threw their famous bombshell in the racing world by produeing ears which showed a startling

ra.vanee in suspension and roadholding qualities. The principal objects of this article are to examine the Methods by which they achieved this advance and to at telnpt to determine the reasons underlying their departures from conventional practice, and so to hazard a prophecy as to the future trend of sportscar suspension.

Dynamically, the advance was accomplished by closer control Of road-wheel movements and by ensuring tmdersteer.

Structurally, it was obtained by using independent front suspension, employing soft springs in cOnjwiction With a stiff frame, mounting the final drive On the frame, and making changes in the disposition of the main masses.

The use of i.f.s. confers a number of advantages. There is Softie S;IVillg of unsprung weight compared with an axle system, but this is not the major advantage,and is not as great as is often supposed. A large proportion of the unsprung weight at the front is formed by the two-wheel, hub, and brake assemblies, together with their steering arms and pivots. These are approximately the same in both systems. Against the saving of the weight of the axle beam must be put the unsprung portion of the suspension links. The biggest saving is achieved by dispensing with 1-elliptie leaf !springs. The unsprung weight of a coil is about onethird that of the corresponding leaf spring, whilst the unsprung weight of a torsion bar is virtually nil.

By far the most. important advantage of i.f.s. is its contribution to understeer. When a tyre is exposed to a lateral force, as when cornering. it tends to drift (without skiddin?.!) in the direction of that force. rnit, angle ‘v ich it ,4 new course makes with the. -ahead is called t he 66 Slip•alWIC:' 'I' in slit -iingle for a given lateral force depends upon various factors ----inflation pressure, loadiiig, and the angle of the wheel to the vertical. I Tnderinflation, overand under-loading, and increase in the angle to the vertical all

increase the slip-angle. Now the directional stability of a car depends mainly upon the relative slip-angles, front mid rear. (Steering and suspension geometry may also affect it.) If the slip-angle at the rear is the greater, the car will tend to turn towards any lateral force. ('entrifugal force -will then be added to tile lateral force, the car will turn still more, and a vicious circle will be estiddished. This is oversteer. If the slip-angle at the front is the greater, the car will tend to turn away from any lateral force. Centrifugal force will then oppose the lateral force, a balance will be reached and the car will be directionally stable. This is tmdersteer.

Since centrifugal force inereases Is the square of the speed, oversteer increases in the same way, and an oversteering car becomes rapidly more directionally Unstable the faster it travels. Consequently, understecr is a property essential in very fast cars.

A potent method of obtaining understeer is to arrange for a large transfer of weight between the front wheels under the influence of lateral forces, with resultant overloading of the outer tyre and underloading of the inner, so producing U large slip-angle at the limit.

When centrifugal forcc acts on a car it transfers weight, from the inner to the outer wheels to a total extent which depends upon the magnitude of that Writ', upon the height of the c. or g. and Intim the track (if the car. The weight transfer between each purr or wheels (front or rear) is made up of two cornponeats, naincly, a share of the moment compelling the sprung portion of the car to roll, and a share of the moment tendho, to upset the misprung portiins or the car.

The magnitude of the ills III causho, roll depends upon the magnitude of ('(lit ifugal force and the distance (if the e. it g. of the sprung portion above the roll axis (that is. the line joining the roll centres of the film! :trid mutt ,“IspcIP:ions). This moment is transmitted to t he wheels through the springs as weight_ transn•r. and the share transmitted front or rear depends upon the relative roll resistances of the front and IV:11'slusja'ulsiIIlis flit WV:tt er the roll resistance at the front the greater the weight transfer due to this component, provided always that the frailly is torsionally rigid. The share. front or rear, of the moment tending to upset the unsprung portiIals of the car clejiends direetly upon the static loading and the height, of the roll eentre at that particular end. The total weight transfer at. cue') end depends on the stiuui itt HICSe 1W0 COMpOlWati.

By far the most important advantage of i.f.s. is its contribution to understeer. I.f.s. gives increased resistance to roll for the same hardness of spring, compared with an axle system, and therefore tends to increase Nveight transfer at the front and produce understeer. A front-axle system offers less resistance to roll than to bump, whilst an independent system offers equal resistance to both. This may be a little difficult to appreciate at first because it is the conventional system which behaves in a complicated dual way and the newer independent_ system which acts more kVit h a conventional axle, for a given level of a front wheel relative to the franas, tie spring on the same side may be in any one of a series of different degrees of compression, depending upon the level of the other end of the axle. In toll, owing to " lowering " of the opposite end of the axle relative to the frame, an outer front. spring is less compressed for a given " rise " of its wheel than it is in

bump," since in the latter case the inner end of the axle remains at its normal level relat ive to the filmic. With independent suspension, for a given level of a wheel relative to the frame the spring on the same side can only be in one particular degree of compression. Consequently, the resistance to a given amount of deflection of that wheel is always the same whether the motion is roll or bump. This increased resistance to roll is slightly offset by the lowering of the roll axis at the front with most types of i.f.s., as will be discussed later, but, there is always a considerable net_ gain in resist anCe 1.0 roll and in weight transfer. A third advantage is that in

all independent systems the path of the steering pivot is positively controlled and the easier angle (except on the Cord) remaiu i,. colistant under spring deflection and brake torque, however soft the spring, so that directional stability is not affected.

These three points—decreased unsprung weight, positive control of the path of the steering pivot, and increased resistance to roll—as well as being advantageous in themselves, allow softer springs to be employed without the disadvantages which attend their use on a conventional front axle. A further advantage of i.f.s. is that there is. no tendency for front axle " tramp " to occur, with the axle oscillating violently in a see-saw manner across the front springs. (Provided the front end of the frame is torsionally stiff. If it isn't, the whole independent suspension asseml;ly may oscillate.)

There are a very large number of different types of i.f.s. They may be classified broadly into three groups. according to the type of wheel movement which they show when viewed from the front.

The first is the "vertical," in which the wheels remain vertical to the road in bump, but lean with the car in roll, e.g., front suspensions of Lancia, Cord, Porsche, Vauxhall, G.P. MereMes, and D.K.W. types. In the first three examples the track remains constant in bump, but in the last two there is a certain amount of side scrub of the tyres. The seeond group comprises the " wishbone " systems, in which the wheels are carried on transverse links, usually of wish-bone form, the upper shorter than the lower, so that the lower parts

of the wheels tilt outwards in bump to maintain constant track. In roll the outer wheel remains almost at right angles to the road, only leaning outwards slightly. Examples of this type are Rolls-Royce and all cars of the General Motors and Chrysler groups, with coil springs ; Lagonda and Citroen, with torsion bars ; and Humber, earlier B.M.W., and Delage, with leaf springs, a transverse spring replacing the lower links.

The third group comprises the " swingaxle " systems, in which the wheels are carried on half-axles pivoted close to the centre line of the car, and swing in arcs whose radius is nearly half the track. The wheels tilt inwards considerably in bump, and slightly inwards in roll to a degree depending upon the distance apart of their pivots. This is the least popular group. Examples are the f.w.d. Derby and the L.M.B. The majority of independent rear suspensions are "swingaxle," notably the original G.P. AutoUnions.

The three groups all attain in only slightly varying measures the basic advantages of i.f.s. explained above, but they differ considerably in their secondary properties. It was, therefore, the secondary properties which determined the German designers' choices.

In all secondary properties the " vertical " group represents one extreme, the " swing-axle " the other, and the " wishbone ' a compromise which can be modified in either direction by adjusting the proportions of the upper and lower links.

(1) The attitude to the road surface which the front wheels assume in cornering greatly affects their lateral drift (` slip-angle ") and therefore the overor under-steering properties of the car as a whole. In the " vertical " group the wheels lean outwards with the car to the full extent of the roll. Drift is at a maximum, and there is a marked understeering tendency. In the " swing-axle " group the position is completely reversed. The " wish-bone " group is intermediate.

(2) The height above the road of the centre about which the front end of the car rolls, and consequently the liability of the system to roll and its tendency to transfer weight, also vary in the three groups. Other things being constant, the higher the roll centre at the front the greater the weight transfer between the front wheels (and therefore the greater the understeering tendency). This is because lowering the roll centre at -the front increases the leverage of the force tending to upset the sprung portion of the car by only about half the amount by which the leverage of the force tending to upset the unsprung portion of the front end is diminished, since the c. of g. of the sprung portion of the car is about half-way along the wheelbase.

This is most readily illustrated by rough diagrams and crude numerical examples. Assuming front and rear spring periodieities to be approximately the same, in accordance with current practice, then load/rate ratios must be the same, and therefore the ratio of resistance to roll front/rear will be as the ratio of static loading front/rear. Assuming—

Track ... .,. = 50"

Weight of car ... ... =

2,000 lb. Ratio of loading front/

rear ... ... ... = 50/50 Height of roll centre at

rear ... ... ... = 10" Height of roll centre at

front ... ... = 10" Height of c. of g. of

sprung portion ... = 25" Then, if u=0.5, and the car is cornered on the point of skidding (see A in the accompanying fig.)—

Overturning moment on front wheels is 2,000 lb. x0.5 x /5" +1,000 lb. x0.5 x 10" 2

=12,500 lb/in. Weight transfer at front is-7 12,500 lb./in . _ 250 lb./wheel. 50"

=50% transfer.

Overturning moment on rear wheels is exactly the same. But if Height of roll centre at front • . • =0" Overturning moment on front wheels is (see B in the accompanying lig.)

2,000 lb. x0.5 x 20" +1,000 lb. x0.5 x 0"

---10,000 lb/in. Weight transfer at front is

10,000 lb/in. —200 lb./wheel 50"

=40% transfer. Overturning moment on rear wheels is 2,000 lb. x 0.5 x20"+1000 +1,000 lb. x0.5 x 10" 2

=15,000 lb/in. Weight transfer at rear is 115,000 lb./in' — 300 lb., ?? I ice I 50"

=60% transfer. For understeer it is, therefore, desirable to have as small a difference as possible in the heights of the roll centres, front and rear. (In this example the unsprung masses have been assumed to be nil, but the error introduced is small, and the principle is not affecte(l.) The position of the roll centre for any system can be determined by super 2 imposing drawings of it in varying degrees of roll. (In many systems the motion of the frame in roll is not pure rotation about a point, but includes some bodily lateral movement, and the roll centre is then not a true point but a small zone.) In the " vertical " group the centre is approximately at road level, its exact position being determined by the type of linkage and the rate characteristics of the springs, but these differences affect its position surprisingly little. Even in the " wish-bone " group it is almost exactly

at road level if constant track is maintained. In the " swing-axle " group, however, it is at the intersection of the lines which join the bottoms of the tyres to their half-axle pivots, i.e., high—usually above hub level.

(3) Side-scrub of the tyres in bump takes Place to a considerable extent in the "swing-axle " group, but is 'absent in the "wish-bone group, and usually absent in the "vertical."

Note that these last two properties are intimately connected, and that any expedient which decreases lateral tyre scrub lowers the roll centre and vice versa. (4) When a rapidly rotating wheel is tilted, gyroscopic action introduces a force which tends to oppose tilting. Therefore, in suspension systems where the wheels are tilted in bump, the inertia of the mass of the wheel to acceleration upwards is reinforced by the gyroscopic resistance to tilt, and, in effect, the unsprung weight is increased. In addition, gyroscopic action in bump tends to produce a movement of the wheel about the king-pin axis, which affects steering. (These effects are obviously greatest in the "swing-axle "group.) The G.I'., 11creedes and Auto-Unions from 1934 to 1939 all used " vertical " types of i.f.s., each marque adhering to, and developing, its own system. Mercedes used pairs of short transverse links of equal length, working against coil springs, which were disposed horizontally in the earlier examples, and vertically in the later. The shortness of the links must have given rise to considerable lateral tyre scrub, but, according to Peter Berthon, wear from this action was negligible in comparison with the rapid

wear from other causes. This tyre scrub may have been embodied in order to provide some frictional spring damping in addition to the hydraulic damping of the shock-absorbers.

Auto-Union used the Porsche system, with pairs of fore and aft arms of equal length working against transverse torsion bars. In the later examples the arms were shortened, as some trouble had been experienced with distortion under cornering loads—presumably twisting, since, with a co-efficient of friction less than unity, the horizontal load would always be less than the vertical.

Since the considerable gyroscopic effects to which it is prone at high speeds ruled out the " swing-axle " system, Mercedes and Auto-Union were compelled to use either "wish-bone " or " vertical " systems at the front. But, as will be apparent from the preceding numerical examples, when the front roll centre is nearly at road level, as it is in both these systems, weight transfer at the front is Almost entirely due to roll. Situ.e both these cars were laid out with wide track, very low build, and rear suspensions having good roll resistance, their tendency to roll and to transfer weight a t the front was inevitably small. It was, therefore, necessary to increase the drift angles at the front by some other means. Adopting " vertical " systems and so allowing the front wheels to lean with the car in roll achieved this. The " wish-bone " layout here had all the disadvantages, tilting the wheels on bump, having a roll centre on road level, and holding the outer front wheel (the more heavily loaded) vertical in roll.

It is interesting to note that E. 11..A. borrowed the Porsche suspension complete, and that the I A-litre, 8-cylinder Alfa-Romeo used a somewhat similar " vertical " system with fore and aft links. The same considerations will apply to the selection of the front suspension layout of any low-built sports car, and %\It strongly favour a "vertical " i.f.s. syst With semi-sporting cars of higher built and lower maximum speed, "wish-bone systems have an advantage. The great,r susceptibility of such cars to roll would give unduly large front slip-angles with

vertical " types of i.f.s., owing to the combination of greater weight transfer due to roll, with excessive lean of the front wheels. "Wish-bone" systems correct this by removing the second factor (as far as the more heavily loaded outer wheel is concerned). The gyrose(ii)ie effects of the corresponding tilt of the wheels in bump is unimportant at moderate speeds. But as the constant tendencies in automobile -evolution are towards lowering the centre of gravity, cutting down roll by expedients such as anti-roll bars, and increasing the maximum speed (e.g., 1941-42 American ears), " wish-bone " systems are likely to decrease in popularity amongst fast cars in favour of "vertical," probably of the Porsche type. The considerations which affected the layouts of the rear suspension of the G. P. Mercedes and Auto-Unions were very different from those operating at the front. The major considerations were the elimination of the shift of weight from one rear wheel to the other under drive

torque (with loss of adhesion of the lightened wheel), and the reduction of unsprung weight. It is difficult to say which was the more important. Happily, both considerations were satisfied by mounting the final drive on the frame.

The 1934 G.P. Mercedes-Benz used a De Dion rear axle acting against a transverse leaf spring, the beam being a forging located against lateral movement by some suitable device, possibly a pair of triangular links meeting at a ball joint, as on. the " Grosser " Mercedes. On later models the forging was replaced by a tubular axle acting against longitudinally placed torsion bars, and located laterally in a slotted projection from the back of the differential housing. The axle was located longitudinally by radius arms running forward from the brake back plates to the sides of the frame.

The 1934 Auto-Union used tubular half-axles, swingiuu on the different ial housitw, actin:_! :1:2;titist a transverse tear SIIIe.;lad stayed hy 10111gitlidirial radius arms as on the lerceiles. By 19:V7 this system had been replaced by a pe Dion layout with torsion bars, very similar to the Mercedes of the same date. The choice between axle Versus independent suspension at the rear was on a different basis from the same choice for the front in important respects. As regards saving of unsprung weight the positions were closely comparable, front and rear, the independent system showing little gain. (Coloparison of I he De Dion axles of the later :1itto-I 'ilions with the swing-axles of their predecessors shows that unsprung weight Nvas increased only by half the weight of the tubular axle plus the weight of the two outer universals. Radius arms were necessary in each case. This increase was. in practice, offset by the substitution of torsion bars for the leaf spring.)

'Positive control of the path of the steering pivot, of course, did not enter into the question, nor, owing to the point of action of the springs close to the wheels, was there any danger of " tramp" at the rear axle. Increased resistance to roll was not an advantage of i.r.s. Roll resistance at the rear is not automatically increased by the use of an independent system, as on a rear axle it is easy to arrange the point of action of the spring so close to the wheel that the degree of spring compression for a given amount of wheel deflection is virtually unaffected by the slope of the axle and is, therefore, equal in roll and bump, as with independent suspension. The roll resistance of a rear suspension system, axle or independent, is normally determined by the height of its roll centre.

The roll centre of an axle layout is at the point at which the axle is anchored to the frame against lateral movement. Obviously. since the corresponding point On the frame does not move laterally in roll, it must be the centre about which the frame rolls. (Where the axle is anchored to the frame not at a point but through some linkage—e.g., a Panhard rod the latter is laid out so that some particular point on the centre line of the axle is always in the centre line of the frame. This point on the axle is then the roll eent re. If, as is often the case, the condition at which the locating device aims is only approximately fulfilled, the roll centre will not be a true point.) With an axle, t herefore, the designer can readily lix his roll centre as high or as low as he likes.

The choice between suspension systems at the rear turned mainly on height of roll eentre and on attitude of wheels to road. The roll centre had to be neither SO low as to cause excessive roll, nor so high as to produce oversteer by increasing weight transfer at the rear at the expense of the front. The attitude of the wheels to the road had to he vertical in roll to avoid a large slip-angle, and as nearly vertical as possible in bump to avoid gyroseopie effects.

" Vertical " systems were ruled Out on both counts, their roll centres being at road level and their wheels leaning with the car in roll, the combination producing gross oversteer. '` Wish-bone " systems also suffered front havite, their roll centres at road level, and although the attitude of their wheels in roll is better, they arc subject to gyroscopic effects in bump. This lull only the " swing-axle " and De Dion systems for consideration.

The roll centre of a system with halfaxles pivoted at the level of the differential axis tends to he too high (it is above the level of the vivots), hilt. the resultant excessive wei?dit transfer at the rear is partly offset by the reversed camber of the wheels on roll, which reduces their' slip-angle for a given deviation from normal load. Reduced weight transfer at the front, however, still occurs. In addition, such a system suffers from severe tyre scrub and gyroscopic effects. The latter are said to have caused its ahaialonment by Auto-Union. (The roll centre of a " swing-axle " system can be lowered by pivoting the half-axles below the level of the differential axis. but this entails the use of two universals each side, so sacrificing the simplicity of the layout. It also gives poor ground clearance under the pivots. Even if by this expedient the roll centre is brought down almost to road level, tyre scrub and gyroscopic effects though reduced, remain always greater than with an axle, since each wheel swings in an arc whose radius is only about half the track, whereas with an axle the radius of the arc described in bump is the full track.)

Since a De Dion axle can have its roll centre at any desired height and maintains its wheels vertical in roll and nearly vertical in bump (a 6-in, bump with 56-in. track means an inclination of only 5'), it fulfils most closely the desiderata for a rear suspension system stated above. It was, for these reasons, the final choice of both Mercedes and Auto-Union.

The considerations which dictated the rear suspension layouts of the German G.P. cars also operate with sports cars, but with slightly different relative values owing to much lower power/weight ratios. Elimination of loss of adhesion of the offside rear wheel at high torques is less important than reduction of unsprung weight, although the former is a considerable advantage under slippery conditions and will probably obviate, even on the fastest sports cars, the necessity for a limited-slip differential—a device which is not an unmixed blessing. The final drive must for both these reasons be mounted on the frame. The height of the roll centre and the attitude of the wheels in roll and bump

are again the determining factors in choosing the type of rear suspension, although, with a higher centre of gravity, roll tends to be greater, and a fairly high roll centre can be employed without too great a disparity between weight shifts, front and rear, in roll, since, as shown in the numerical example given previously, weight transfer at the front, when the front roll centre is at road level, depends on the rolling moment. Gyroscopic effects, too, are less important on sports cars, since speeds are lower. The "swingaxle " is therefore not entirely ruled out, but as with " wish-bone " at the front, lower centres of gravity and higher speeds will tend to make it progressively more disadvantageous. In addition, its severe tyre scrub, if soft springs are used, is a considerable drawback for normal use. Within five years of the resumption of car production the majority of expensive cars will have De Dion rear axles. The Americans are likely to be early in this field—as they were with i.f.s. (The use of coil springs at the back on Buick, Oldsmobile and certain Chrysler models suggests that they feel the time is ripe for radical changes in rear suspension.) In passing, it is worth noting that if fore-and-aft radius arms, rigidly attached, are used to locate a rear axle, one arm will be twisted when one wheel rises relative to the frame, making the suspension stiffer than when both wheels rise together. If a torsionally stiff axle beam is used, this effect will be severe. It is

undesirable, as the springing will be either too hard in the first case, or too soft in the second. The effect is exaggerated in roll, when the axle will act as an anti-roll bar, with increased transfer of weight outwards between the rear wheels, tending to produce oversteer. The remedy is to make one radius arm free to rotate on the axle, or to use an axle section which, though stiff in bending, is weak in torsion. Citroen tried the first solution on the rear axles of his earlier front-drive models, but quickly discarded it in favour of the second, using a cruciform axle section.

If it is necessary to increase the resistance to roll of a suspension system, it is better to put the anti-roll bar at the front, when there will be increased transfer of weight outwards between the front wheels, with correspondingly decreased transfer between the rear, so assisting in obtaining understeer. The front suspension is likely, in any case, to be softer, and its resistance to roll therefore less, than the rear, as by such an arrangement fore and aft pitch is automatically damped out quickly. For these reasons the use of anti-roll bars at the front on touring and semi-sporting cars is likely to increase. The Americans have already made a start. The last innovation introduced by the German G.P. cars to improve suspension and roadholding was redistribution along the chassis of the main masses, which were pushed towards the ends. This was quite evident on the Mercedes, but it is doubtful Continued on page 86

SOME SPECULATIONS —continued from page 80

to what extent it was done on the AutoUnion. It was the manifestation of a tendency which was already apparent on touring ears, where it had produced the forward engine mounting, but it had not at that date been seen on racing cars. On the Mercedes the radiator was mounted forward of the front hubs, and the engine came immediately behind it, in marked contrast to contemporary racing practice. In spite of this, 60 per cent. of the loaded weight of the vehicle was carried on the rear axle, showing that the remainder of the main masses were mounted well towards the back of the chassis, notably the gearbox, which was moved from its conventional position amidships and combined with the final drive. Moving the main masses to the ends of the chassis increased its moment of inertia, and thereby its resistance to disturbances in both the vertical and horizontal planes, thus giving a "flatter" ride and greater directional stability. This distribution of masses is now almost universal on touring cars and is steadily increasing in popularity on sports cars. [Pushing forward the engine on normal cars was first done, of course, to give increased passenger space. —ED.]

From this examination of the German G.P. racing cars the pattern of the postwar sports car clearly appears. In the larger and more expensive classes it will closely resemble in suspension arrangements and in general layout the G.P. Mercedes, with stiff frame, soft springs, " vertical " independent front suspension, De Dion rear axle, forward engine mounting and gearbox in unit with the final drive. The fast touring car may differ in having " wish-bone " i.f.s., and " swing-axle " rear suspension. The smaller sports car has to scale down its suspension movements to suit its narrower track and shorter wheelbase. Consequently, its necessity to use i.f.s. to control king-pin attitude is less pressing,

though still essential if a comfortable ride is to be combined with understeer. For this, and economic reasons (technical advances usually spread down from the more expensive models), the smaller cars will prove slower to change over to i.f.s. At the rear stiffer springs, which permit them to profit less from reduced unsprung weight, lower torques, and considerations of cost, will conspire to delay for at least ten years their adoption of the De Dion axle, and, therefore, of gearbox in unit with the differential. In the meantime, many of them will have gone over to front drive. *

* WANTED

The Editor will be pleased to hear from anyone who has for disposal copies of

Monte Carlo Rally" and "Motor Cycle Reminiscences," by " Ixion," and a model Pfe, Alfa-Romeo car or other car models.