While the evidence summarized in the previous chapter did suggest that herring gulls and gannets managed to home without any highly developed ability to select the home direction, there was no compelling reason to apply this explanation to all birds. Suspecting that other kinds of birds might well display superior powers of navigation, Harold B. Hitchcock and I turned to homing pigeons in an attempt to learn something about their navigation by tracing their flight paths from small airplanes. Hitchcock's pigeons performed better than mine, and therefore Fig. 16 shows some of the longer flight paths which he succeeded in following. It is apparent that these pigeons deviated considerably from the direct path between release point and home, but they did better than one would expect from the molecular theory of bird orientation.

In the early 1950s two important advances were achieved in the explanation of bird navigation, one by Geoffrey V. T. Matthews at Cambridge University, in England, the other by Gustav Kramer at Wilhelmshaven, Germany. Matthews worked mainly by observing and analyzing the initial headings of birds in homing experiments. Kramer's experiments (to be discussed later) included not only observation of the initial headings of released homing pigeons but also study of the migration restlessness (described in Chapter 3) of a few selected, handraised, and caged wild birds.

For many years pigeon racers had maintained that on release homing pigeons often would fly straight off toward home. But in most casess racing pigeons are released in large flocks containing many birds experienced withh the particular release point and the homeward route. A reasonable explanation, in the absence of other information, was that these experienced birds remembered the correct route and were followed by the others. In many homing experiments the immediate surroundings at the release point sometimes seemed to have an effect on the bird's choice of direction for its first flight. For instance, sea birds released inland tended to head for any body of water, while pigeons were likely to display some interest in towns or farms.

Matthews' Observations of Initial Headings

Matthews made a special effort to select release points at the centers of large open spaces of level terrain with a clear view in all directions. Every effort was made to avoid local features of topography that might influence the direction in which the birds would fly when first set free. To guard against bias it was necessary, of course, to vary the direction in which the birds' heads were pointed at the actual moment of release. This pointing was done in an irregular fashion, perhaps with the first bird headed north, the next southwest, the third east, and so forth.

In his first experiments Matthews trained pigeons by releasing them at increasingly distant points that all lay approximately on a single line. This training direction extended north northwest from Cambridge. As in all properly conducted homing experiments, the birds were transported in closed boxes, usually by automobile or light truck, and they had no opportunity to see anything of the countryside. When thus released in unknown te.ritory in an unknown direction from home, many of the pigeons headed initially in their customary southward homing direction. These birds, it was apparent, had established a preferred direction of flight which they chose regardless of the release point. When taken to a new release point on the training line, several miles farther north than their previous starting point, the pigeons again displayed a clear ability to fly south-toward Cambridge.

Naturally, Matthews wanted to find out on what basis this choice was made. By releasing pigeons under a variety of weather conditions, he made the important discovery that the initial headings were far less consistent when the sky was heavily enough overcast to hide the sun. While there was the usual variahility from day to day, and from one pigeon to another, the choice of an appropriate direction seemed to be greatly impeded, if not rendered altogether impossible, when the pigeon could not see the sun. It had long ago been suggested that in pigeon races, which always start early in the morning, the birds inight take advantage of the sun's position to deterinine the appropriate homeward direction in which to fly. But it seemed incredible that a bird might use the sun as a "compass" to determine directions at any time of day. Matthews was the first to consider this possibility seriously enough to carry out esperijnents that tested it directly.

In one series of releases along in the training diicction north northwest from the pigeon's home aloft near Cambridge the releases were made always iii the middle of the day when the sun was nearly at its highest position in the sky. It was thus approxiiiately south of the birds when they made their ini( iA choices of flight direction. The simplest esplanai ion for these directional choices would be that the I)ird had learned to fly toward the sun. But these pi;cons did not merely start in the correct direction; niany of them also reached home at relatively high speeds. A simple tendency to head toward the sun would have caused them to turn more and more to the west as the afternoon wore on, and by evening they would have been flying almost directly away from home.

Matthews investigated the matter further by experimenting with pigeons which had been released several times at increasing distances up to seventyeight miles along the training line, always at approximately midday. Dividing the flock into three groups, he transported them forty-nine miles farther (that is, 127 miles from home) in the same direction. One group was now released early in the morning, another late in the afternoon, and a control group in the middle of the day as usual. On the average the experimentals were released six hours earlier or later than the controls. The result was that all three groups chose the homeward direction with equal accuracy. It made no difference that the sun was roughly ninety degrees east or west of the position where they might have come to expect it. The results of this important experiment are illustrated in Fig. 17, which shows not only the similarity of the headings when birds were released early in the morning, at midday, and late in the afternoon, but also the general accuracy of these homeward headings.

After Matthews had completed the experiments just described, he did not rest content with the demonstration that his pigeons could use sun-position for headings. He continued and extended the same basic type of experiment and soon found that some groups of pigeons performed in a still more interesting manner. After considerable experience at distances up to fifty miles or more, along a single training line, they were released in unknown territory in some quite different direction. In the most important experiments the training direction was again north northwest, but the crucial tests were made at release points to the south or west. Again there was considerable variability, and a few of the birds flew roughly south as they had done on previous occasions when released along the training line. Others scattered in many other directions. But a considerable majority flew off more or less toward home. This was true whether home was north or east. After a number of experiments had been completed, it became clear that many of the pigeons could select approximately the homeward direction whether or not it coincided with that of the previous training flights. Examples of several experiments of this type are shown in Fig. 18. Despite the considerable variability, the choices made by the great majority of birds lay far too close to the home direction to be explained by random scattering, or by any of the explanations that appeared applicable to the gannets and other homing experiments prior to those of Matthews.

Fig. 17, Initial headings of pigeons trained by several releases at midday. Each release was farther north northwest from the home loft, the release previous to this one having been at 78 miles. At left, the initial headings of birds released at 127 miles (in totally strange territory) around noon. At right, the headings of others from the same group released at the same place in early morning and late afternoon. The length of each line is proportional to the number of pigeons starting in the direction indicated; the shortest lines signify one bird. Small arrows show the directions these birds would be expected to take if they had flown at the same angle to the sun as in the training flights.

In spite of the definite demonstration that pigeons could establish homeward orientation within the first few minutes after release in unknown territory, the interpretation of such initial headings is complicated by the fact that in every experiment at least two types of orientation must be considered as distinct possibilities. Are the birds only choosing a roughly constant direction (in the case of pigeons usually a training direction)? Or will they choose the true home direction regardless of which way they may have been displaced into the unknown surroundings of the release point?

If a bird is taken in the direction opposite to a preferred flight direction, accurate homeward headings do not establish which type of orientation is involved. Other experimenters with pigeons have found that their birds exhibit a constant directional tendency in their initial headings independent of the direction they have been carried from home. For example, pigeons trained to home to Wilhelmshaven near the northern coast of Germany, where most of Kramer's experiments were conducted, show a northward tendency in their initial headings. Only after releases in two or more different directions can one determine with reasonable confidence whether a uni-  directional tendency is at work or true homeward orientation.

The Initial Headings of Manx Shearwaters

In addition to his work with domestic homing pigeons, Matthews also experimented with Manx shearwaters, birds which had already been shown to have a most impressive homing ability. It was Manx shearwaters which returned from Venice and Boston to their nesting island off the coast of Wales at average speeds of 265 and 244 miles per day. Matthews transported shearwaters from the same island to various points inland in England where the terrain was relatively uniform and an observer on a good vantage point (such as the library tower at Cambridge University) could follow the birds with binoculars for considerable distances. Again there was variability, some shearwaters choosing directions far removed from the correct homeward path. But, on the average, shearwaters performed as well or better than the best homing pigeons, and a tabulation of results from several inland release points in England showed beyond doubt that most of the shearwaters were choosing approximately the correct homeward direction and were not scattering in anything approaching a random pattern.

These experiments obviously carried the whole problem of homing in birds into a new phase. Other investigators tried essentially the same methods on other species, but no other bird has yet been found which shows as consistently accurate initial headings as the homing pigeons, used by Matthews and Kramer, or the Manx shearwater. Other not-so-select strains of homing pigeons give results that are far less convincing. Nevertheless, the fact that some birds can consistently choose approximately the correct homeward direction, and demonstrate the correctness of this choice in their initial headings, means that we are dealing with an ability to orient which makes itself manifest within a very short time after release. No one can yet say with certainty whether it is only a few species of wild birds, and only the best strains of homing pigeon, that possess this ability, or whether it is widespread among birds but has only been demonstrated in a few favorable cases.

Kramer's Experiments with Orientation Cages

At about the time when Matthews was conducting the experiments described, Gustav Kramer and his associates were experimenting with orientation exhibited under more strictly controlled conditions in experimental orientation cages. In his first investigations Kramer used starlings which had been raised by hand from an early age, and were very tame. They were placed in circular cages provided with perches both at the center and also around the edges. As had been discovered many years earlier, such migratory birds exhibit a pronounced restlessness at the season when they would normally undertake long migratory flights. As explained at the end of Chapter 3, this migratory restlessness occurs primarily at night, when many small birds migrate, but Kramer's clearest experiments were carried out in daylight, employing a few well-tamed starlings, which are diurnal migrants.

The starlings were placed in a circular cage which was carefully built so as to be completely symmetrical; its inner surface had an identical appearance when the bird faced in any direction. The observer lay on his back on the floor and watched the birds  from below through the transparent plastic floor of the orientation cage. One tame and co-operative starling tended in the spring to fly back and forth from the central perch to the edge of the cage primarily in a northwesterly direction. This direction corresponded roughly to the normal northeasterly spring migration of starlings in the area where it had been taken from its nest in its first few days of life.

On what basis did this starling make its directional choice? In addition to being symmetrical, the orientation cage was arranged so that it could be rotated, and when it was turned from time to time during the period of migratory restlessness the starling continued to head in the same general direction, regardless of which wall of its cage had to be approached in order to head northwest from the central perch. The next step was to surround the orientation cage with a uniform, opaque fence or shield, which also could be rotated about the center of the cage. This screen cut off any view of the surrounding landmarks, but allowed the starling to see most of the sky. Again it made no difference how the screen was rotated. Clearly the bird was not responding to local landmarks. Kramer also established that the bird and orientation cage could be moved about from point to point in the vicinity of Wilhelmshaven, without occasioning any essential change in the directional tendency.

What aspect of the sky provided the directional information underlying these choices of the correct direction for fall migration? One possible directional cue was the sun. For the decisive experiment Kramer enclosed the orientation cage in a further opaque structure which excluded all view of the outside world except what the bird could see through six large windows. This starling continued to show the same northwesterly heading when it could see the sun and blue sky through one or more of these windows. The next step was to equip the windows with shutters whose inner surfaces had large plane mirrors, as shown in Fig. Iq. These shutters were so arranged that the light reflected into the window by the mirror came from ninety degrees to the right or left of the window itself. In other words, looking out the window, the bird saw reflected in the mirror an area of sky ninety degrees to one side of the sky directly beyond the window. The starling continued to show strong directional preferences, but now the bird maintained the same angle relative to the light reflected from the mirrors as it bad to the normal sunlight. Thus, when the mirrors showed the sky ninety degrees to the right of what normally would have been visible through each window, the starling headed approximately ninety degrees to the right of its previous heading. The opposite result occurred when the windows were so arranged that the sunlight reached the bird from ninety degrees to the left of its normal angle of incidence.

When the sky was completely overcast the headings were virtually random, but when the sun became visible a few hours later the bird again chose approximately the northwest direction for its headings. Kramer thus showed that one particularly cooperative starling was able to choose a consistent direction in his circular orientation cage, and, furthermore, that this direction was determined with reference to the sun. The correspondence between this experiment and Matthews' studies of initial flight direction in pigeons was obvious and compelling.

ANIMAL NAVIGATION
(Extract from Animal Navigation by R. M. Lockley Science Series, 1967)

It is now known that many if not most of the non-migratory birds lack - presumably because they have never had occasion to exercise it - the innate ability to orientate over long distances possessed by pelagic and oceanic migrants such as the albatrosses, shearwaters, terns, etc. At most the purely sedentary species can get back home only if they are released within or close to their local territories, with some exceptional individuals doing rather better than this. The case of the racing, carrier or homing pigeon is interesting, because, although it has been extensively used in scientific homing tests, it is also by origin a non-migratory bird. It derives ftom. the wild rock-dove, normally resident within a very few miles of its rock caves and cliffs throughout the year - a strictly vegetarian feeder.

Tales of prodigious horming feats of this domesticated dove must be discounted; they may be boasts of imaginative racing pigeon fanciers. The many careful experiments prove that their sense of direction is limited, and that they are lost if transported hundreds of miles out of the area in which they have been trained to race. Like dogs which have been trained to home in a specific direction they will return home rapidly over lengthening distances only if released in strange country along the compass course with which they are familiar. Even so it is usually necessary for pigeons to see the sun (they do not fly at night) so as to orient on the correct bearing for home. Racing pigeon fanciers fear misty weather when they send their trained birds away on competitive races; many are lost then, wandering to strange lofts, or settling down to live with wild rock-doves on cliffs, or gone-wild pigeons on town buildings. They seem to lose heart if they cannot get back to their home loft in the same day; and one pigeon fancier tells me that even if he does eventually recover a lost bird it is seldom much use as a racer afterwards and he will not keep it as a breeder in case it perpetuates its failure in its offspring.

In a series of experiments which I conducted with the Editor of the Racing Pigeon, five untrained young pigeons were released in Wales 250 miles from their home loft in London, and five more from the same loft at Cardiff, i5o miles from London. Although these youngsters had been flying strongly above their London loft, and knew it visually from the air, they had never been sent away on races, but were robust and physically in perfect flying condition. Yet none was ever recovered back home again; only two were reported, some weeks later, at lofts completely off course in the Western Midlands, at least 70 miles from London. This proved that these young domesticated rock doves, like other non-migratory birds, have no innate ability to navigate when displaced so far out of their normal range. They are convenient to use because of their sedentary habit of returning to a communal roost, but not ideal navigators in such experiments.

It is, however, possible to train them, and it was with well trained pigeons that G. V. T. Matthews had some interesting results. By releasing groups of experienced homers farther along their training line at different hours of the day, he found that the position of the sun in the sky at the moment of release made no difference: the majority flew off in the correct direction for home, and only a minority headed in the opposite direction.

The experiments were repeated on a larger scale, with five widely separate release points; this time many of the birds were released in unknown country in a direction opposite to that of the previous training one. The results are shown in Figure 10, and confirm that some, but not all, pigeons are able, after training, to select a homeward direction if released almost anywhere within i5o miles of home. But in misty weather they were quite disoriented.

The homing experiments with oceanic birds quoted above indicate that those species, under test, can correct much longer deviations from their normal range artificially arranged by man, and find their way home over many hundreds of miles of unknown sea and land. On reflection it is clear that such an ability is essential for survival when strong winds blow individuals far off course into regions unfamiliar to them, but how they re-orient so accurately remains a puzzle.

Several years ago G. V. T. Matthews suggested that the day-flying homing pigeon at the moment of release is able to detect the arc through which the sun is travelling, and predict the arc to its highest point (due south in the northern hemisphere) and so obtain its latitude. At the same moment the bird's keen sense of time tells it where the sun ought to be at home (or home time at noon) which will give it its relative longitude. But how precise a bird can be in calculating so fine a distinction we do not know. When man makes such a calculation by the heavenly bodies he has to allow for several, but principally three, variations: for dip, which is the angle of  epression of the visible below the sensible horizon; for parallax, the angle subtended at the heavenly body's centre by the observer and the earth's centre (when a body is at the observer's zenith the angle is 0 degrees); and especially refraction - a ray of light from a celestial body is bent as it passes through layers of increasing density of the earth's atmosphere.

In the northern hemisphere mallard, in homing tests in both England and the United States, seem to be drawn towards the North Star in what Matthews describes as 'nonsense orientation', before they eventually swing round and resume their migration in the normal direction. In eastern North America Grfffin found that common terns headed initially in the opposite direction - south cast; although some of the terns were nesting in that direction, others were breeding north-east of the release point. Even more puzzling is the report that some common and black-headed gulls were able to orient themselves towards the place of capture from a totally enclosed room with no view of the sky.

Matthews's 'North Star mallards' were the subject of several other experiments, continuing as I write this. He 'reset' their biological clocks by keeping them away from daylight in an enclosure, under artificial schedules of light, for several days until some were accustomed to a day six hours ahead, and some to a day six hours behind, the true day. A third group was accustomed to a day twelve hours out of phase. When he released in daylight all three groups, together with a control group which had been living under natural daylight, the controls headed north-west as expected. Those with clocks reset six hours ahead flew south-west, those with clocks six hours slow flew north-east, and those with clocks twelve hours out of phase, flew in the opposite direction - south-east. Thus the experiment with sandhoppers  was duplicated with ducks, with the same result. But when Matthews released the same three groups of ducks with clocks reset as before, but under the stars on a clear night, all the mallard oriented northwest, as did the controls!

Man has found that the stars (but not the planets), fixed in their courses at an immense distance from Earth, are more reliable as navigation guides than the great sun itself. So, too, it seems, sidereal pattern and time have a greater influence on animal navigation than the sun, at least where a species, like the mallard, migrates by both day and night.

There is an acceptable theory that in general migrating birds follow the shortest possible line between summer and winter quarters, a line which can only be drawn accurately on a globe atlas. Any 'straight' line extended far enough over and parallel to the surface of the earth will automatically complete a circle of the same length as the equator. Such a line is known as a Great Circle. Plotted on a flat projection of the world, the flight of a bird such as a warbler, migrating from summer quarters in eastern Siberia straight to winter quarters in Kenya, would appear to pass southwards through the Gobi Desert of China, India, and over the western Indian Ocean. In fact its shortest route along the great circle actually begins with a westward flight into Europe, and only gradually southward via the Aral Sea and the Red Sea to Kenya - an all land-route and less perilous for a small insectivorous warbler.