Soaring Machines

By L. Hargrave.

Read before the Royal Society of New South Wales, November 2, 1898

It should be noted that in this article, Hargrave refers to "propellers" which are not airscrews, but are in fact aerofoil shaped wings, designed for soaring, converting the wind's energy into forwards and upwards motion. When Hargrave mentions "drift" he is talking about a concept he developed, taking into account the drag (which he called head resistance) and downwards pull developed by a flying line.

It is long since my diary contained sufficient matter to submit two papers in one year for publication in our Journal, but some late observations are of such a convincing nature as to the truth of the propositions enunciated here on August 4, 1897, that I trust you will permit me to advance the art of soaring another step.

Figures 1, 2, 3, show side and end elevations and plan of two soaring kites that are called M and N. The following table is the type of the inscription that is plainly legible on the photographs but may be indistinct on the zincotypes:-

Soaring kites




4' 11"

4' 1 1/2"


3' 3 3/4"

3' 0"

Projected area of propeller, square inches



Projected area of end surfaces, square inches



Total area, square feet



Weight, pounds



Weight per square foot, pounds



Angle of propeller

-4 1/2 degrees

-3 degrees

Both kites have repeatedly soared in wind with a velocity of ten to fifteen miles.

M and N differ in several ways from the vulcanite soaring kite described in the paper of June 1, 1898. The long tin tubes are much stiffer and the propellers are made of redwood. M also has a spring screwed to the front of its propeller so that a trial could be made with the propeller rigid followed immediately by one with the propeller springy. This showed the rigid propeller to be the best.

It is found that vulcanite immersed in boiling water and then bent does not retain permanently the curve imparted to it; neither does steamed wood unless nailed to numerous objectionable ribs. Bent metal is worse than bent wood and weight for weight is more flexible. There appeared no alternative but to work the curve of the propeller out of solid wood; this course produces with some patience the desired article. When the best curve has been decided on, curved wooden propellers will be produced by modern wood working machinery with as much facility as any form of moulding used in architecture.

A further consideration of the horizontal projection of a soaring bird's wing shows that the tip or flat part is approximately half the area of the soaring part.

When the wing is rigidly extended and the soaring part lifting properly; the tip, when in the plane of the true wind, will have the relative wind acting on its upper surface, and be in effect a kite wrong side up.

The position of the tip, that is, whether it incline up or droop down when viewed from front or rear, is a clear indication of whether the bird is soaring in a horizontal or downward blowing wind, or merely being supported by an upward trend of wind.

The albatross and frigate bird show the drooping tip to perfection. Hawks and eagles frequently show the upward bent tip, and when they do so we may safely conclude that any flat object would be up-borne by the wind in their neighbourhood. Birds that circle in calm or nearly calm air, have the wingtips turned up; and if the performance takes place over a hot and dusty plain, the conditions are favourable for the formation of a sand column or "whirley". The bird that soars in a gale has a deeper concavity of wing than one that soars in a moderate breeze, from which I deduce that the velocity of rotation of the vortex must have somwe point of maximum efficiency. In other words, the small vortex cannot attain an infinite velocity, and the large vortex loses its efficiency when its speed of revolution is reduced below a certain point. Each form of soaring wing is evolved by the average velocity of wind in the latitudes frequented by the bird.

A kite (O) was made four feet wide and seven inches in a for and aft direction. Two feet six inches of the middle was shaped to a soaring curve and the rest left flat. The inclination of the flat part to the chord of the soaring part was 6 degrees and unadjustable. A rod with weighted ends and small tail was added. This kite soared several times but was crank athwartships.

It was thought desirable to reject all horizontal surfaces as it appeared that their only use to a bird was to enable it to fly when there was no wind; and as these soaring kites had no motor but the soaring curve, the flat surfaces only increased the drift. At this point the soaring machine develops into a form that has no counterpart in nature. The rod now having no horizontal surface at the ends; could not, by the inertia of the lead weights alone, long retain the propeller at the proper soaring angle. The machine must sooner or later tip either up a down. The rod with loaded ends and cells can but retard the end tipping long enough to show that the propeller is soaring. For these reasons the weight was transferred to a point below the propeller, thus reverting to the methodd of maintaining the equilibrium of the balloon or parachute and which is used by the experimenters with gliding machines.

The situation and aspect of the tail or weather-cock came under consideration, and it was seen that the nearer it was placed to the after edge of the propeller the'" more instantaneously would gravity adjust the propeller to the proper angle. It was also recognized that whatever area is given to the weather-cock, its longest dimension should be vertical. The meteorologists will think this rank heresy.

Lancaster points out that the weather-cock should be vertical only; and as far as I know every aeronautical construction ever made but his, has horizontal tail surface. A moment's consideration should have shown us that when we wish to preserve the angle of incidence by the action of gravity, as all gliding machines do, any horizontal tail must act as a check to the necessary rapid adjustment.

Kites M. and N. were therefore remade as shown in Fig. 4, 5, 6, 7, and assume a strong likeness to Lancaster's "effigy" described in the Engineer 1882, and which I have endeavoured to reproduce in Fig. 8, from his dimensions given in Chanute's "Progress in Flying Machines", page 199. I can well believe that many of Lancaster's "hundreds of effigies" soared in spite of their flat cardboard surfaces, if the stick that extended the wings had some considerable depth and was fastened to the under side of the leading edge; which point is not made clear.

It is also recorded in the same work at page 197, that his explanation in the American Naturalist was so plainly erroneous that he was harshly criticised.

I think that Le Bris in 1867, Mouillard (date unknown), and Lancaster in 1882 all made soaring machines that worked by means of the soaring vortex, although there is no record of their having known or shown that the air in contact with the rear side of the leading edge was at a higher pressure than that on the windward side.

Phillips in 1884 and Montgomery (date unknown) showed that the air at the rear of the front edge of a similar curve to a soaring bird's wing was moving downwards, but both of them just stopped short of finding the high pressure of the vortex.

Lilienthal found that arched surfaces produced a lift slightly to windward of the zenith, my work being published in 1893 being identical with his.

If there are others who have made soaring machines and showed why they soared, I have omitted to mention them through ignorance, but in a matter concerning claim to priority of discovery the credit must go to the man who first publishes his knowledge, and none at all to the one who knew and withheld his information with a view to exploiting humanity.

I have noticed that soarinig is easier in a wind velocity that is increasing than when it is decreasing, and attribute this to volumes of air of high density and velocity driving in under volumes of lower density and velocity: the contiguous surfaces will then cause eddies in the combined mass rotating in the smae direction as the soaring vortex does: that is, the upper part of the eddy moving to windward and the lower part to leeward, one of these would be more readily caught and held by the propeller than when contrary conditions prevail.

Every detail of Kites M and N as remodelled are shown in Figs. 6, 7, and they now contain all the necessary parts of a practicable soaring machine to carry one man, and I expect to hear ere this is in print, that some of the gliding machines on the shore of Lake Michigan have been fitted with soaring curves, the trials of which are certain to be successful.

The observations made on August 31, 1898 are as follows:- Kites M and N to the beach. Very steady east wind, twelve to fourteen miles. No sea to speak of or that might cause large pulsations in the wind. Poles placed close to the water. The waves washed around the two windward pegs. Sand almost level. Rain beating the models down.

Kite M hung by thirteen feet of cord .15" diameter = 23 sq. ins of cord for head resistance.


Kite weighs without 	ballast		1 lb. 12 oz.
			Ballast		3 lb. 1 oz.
  		   Total weight		4 lbs. 13 oz. = 4.81 lbs
    Projected area of propeller 	2.58 square feet.
			   Load		1.86 lbs. per square foot.

When M was loaded with 3 lbs. 1 oz. of lead she hung persistently 7 degrees to windward of a plumb line passing through the after end of the tail and the knot that attaches the hanging cord to the horizontal one at the top of the poles. Sometimes she would swing back till the hanging cord was from one to two degrees out of plumb. The plumb line and weight were sheltered as much as possible from the wind by my arm.

When the kite is drawn about four feet back from the vertical position and released, the hanging cord slacks when the kite has swung about two feet forward, and M soars with a deep bight in the cord to position B (Fig. 9), and then turns and rushes round like a conical pendulum, jerking savagely at the hanging cord in all directions. It then has to be caught as it is impossible to tell what is real soaring and what is impulse derived from elasticity of the poles and cords.

Kite N was then attached to the horizontal cord by a piece of fishing line and loaded with 2 lbs. of lead.

The area of N's propeller is 243 square inches = 1.69 square feet.
		N's weight without lead		1 lb. 0 1/4 oz.
		Lead weights			2 lbs. 0 oz.
						3 lbs. 0 1/4 oz. = 3.016 lbs.
		Weight per square foot = 	1.78 lbs.

Kite N starts from a plumb position and ascends slowly at an angle of about 45 degrees to windward, it did it five or six times in spite of the rain beating it down, and the drift of the hanging string and a light line tied to the weight to keep it from dashing about. Fig. 9 shows a side view of the experiments with M.

It may be thought that it would be more conclusive if the models were allowed perfect freedom. This matter has not escaped consideration, and the reasons for not working with free apparatus at present still hold good. By using the captive method, any amount of skill and patience expended in the manufacture of the soaring machine is amply repaid by its possession and the knowledge that the experiment can be repeated under similar conditions. Whereas if the free method were used, a form that merely wanted a little adjustment to be perfect, would frequently be smashed or lost in the sea without anything remaining to show its defects or lead to rapid improvement.

Of course if I lived in the centre of a sany plain, with numerous assistants to make and repair constructions of my design, certain advantages would accrue, but at present I try to make the utmost use of the facilities at my disposal.

Fig. 10 shows the condition of the air in the neighbourhood of the soaring curve and the following statements may help us to arrive at the exact power developed:-

  1. The hook originates the vortex. 
  2. The diameter of the vortex is determined by the radius of the race
  3. The velocity of rotation is something less than the velocity of the wind or relative wind, and is maintained thereby. If the wind is thirteen miles and the curve advances into the wind at one mile, the relative wind is fourteen mile and the velocity of rotation about 2,600 revolutions per minute.
  4. The air drawn in from the rear of the vortex rises in pressure as the race contracts.
  5. The high pressure air in the race acts on the soaring machine by thrust on the vortex nest.
  6. The vortex cannot increase in diameter or burst because the vacuum at the centre is of the exact tenuity that balances the centrifugal force of the particles of air forming the vortex.
  7. If the head resistance of the soaring machine is decreased by a lull in the wind, the air in the race expands leaving the vortex slightly to leeward, that is practically increasing the radius of the vortex nest, the vortex then increases in diameter and rotates slower, draws in less air past the guide and restores the equilibrium.
  8. Some of the discharge from the race may pass into the dead air to windward of the hook and so over the top of the soaring curve, or if the dead air space is filled up solid with part of the material of the soaring curve the whole discharge is carried under the vortex and may or may not be drawn in again between the vortex and the guide. The discharge cannot mingle with the air of the vortex, as every circumferential particle of its air is held at a fixed distance from the centre by the tenuity of the vacuum.
  9. The lower front quadrant does not add to the head resistance as it is rotating to the leeward nearly as fast as the relative wind.
  10. The after part of the soaring curve if it extends to leeward of the divide acts as an aero- curve.

On October 20, 1898, the wind was about seventeen miles per hour, and it was found that kite N could be loaded with lead to a total weight of 3.6 lbs. on 1,69 square feet = 2.13 lbs. per square foot, and that when so loaded it would rise at an angle of 70 degrees or 80 degrees to windward until it was fifteen feet from the sand, it then got into wind of greater velocity and drifted to leeward. Here I am confronted with a difficulty that at present is unsurmountable. Either the soaring machine must be started from such a height that the weight can be approximately adjusted to the existing wind; or, the weight must automatically adjust the negative angle of the propeller as the wind increases.

Kites O (Figs. 11, 12), P (Fig. 13), Q (Fig.14) have a different system of adjustment and suspension of the weight. A piece of 3/4" tube is secured rigidly to the propeller and nearly parallel to its chord. The connection between the tube and propeller in O and Q is a steel plate 13/16" X 1/16" and long enough to keep the weight at the required distance below the propeller. The weights are lead cylinders 3/4" diameter and about 1 1/2" long. A suficient number are strung on a 1/8" wire. The adjustments of the position of the weight is effected by pushing the string of weights in or out of the tube. The head resistance is thus reduced to that of the edge of the plate plus the end area of the tube.

Kite O has the weather cock attached to the after end of the tube, and is the kite previously mentioned, now remade.

Kite P has the upper side quite flat, the hook is 3" abaft the sharp leading edge of the propeller. The space between the hook and the leading edge of the propeller is solid wood slightly concave, so that there can be no dead air to windward of the hook.

A fringe of silk is glued to the concave side of the propeller on O and P so that it is possible to see that the "divide" is approximately in the position shown in the diagram (Fig. 10).

Kite Q has two propellers superposed at a distance of 8". The ballast tube is 3 1/4" below the under one. This kite shows that a double propeller soaring machine can be ballanced in a fore and aft direction as well as, or better than, the single form.

Return to "The Hargrave Files" or to the AKA home page