On the Reactions to Humidity of Peripatopsis Moseleyi (Wood-Mason)

The species chosen for study was Peripatopsis moseleyi (Wood-Mason). This species occurs in moderate numbers in many Natal forests and is invariably found in damp places, usually underneath old decayed logs of wood or burrowing in their substance. In the laboratory the animals were kept in shallow photographic dishes filled with pieces of decayed wood. These were sprayed with water once a week. The dishes were covered at one end with a glass plate and at the other with perforated zinc sheeting. The gradient of moisture thus produced lessened the danger of fungal infection, which is common in damp cultures.

The alternative chamber described by Gunn & Kennedy (1936) was used to establish the presence and nature of the humidity reaction. In this apparatus the animal is presented with a choice of two humidities. The preference for one or the other may be quantitatively estimated by determining the distribution of animals in the two halves of the chamber when an equilibrium has been reached. The desired humidities were maintained by solutions of potassium hydroxide of suitable concentration.

The nature of the experimental material necessitated certain modifications of the procedure employed by Gunn (1937). Only one animal could be used in each chamber, since a strong tendency to aggregate was evident when several animals were present. Secondly, the usual methods of activation could not be employed; light, as used by Kennedy (1937), was ineffective, while mechanical stirring was undesirable as the animals reacted by ejecting slime in which they frequently became entangled. It was found that activation could be brought about simply by rotating the false floor of the chamber so that the animal was brought back into the central region of the gradient. The direction of rotation was such that the animal was always turned through more than 90°, but never more than 180°. Experiments with cobalt thiocyanate paper (Solomon, 1945) as a humidity indicator showed that the effect of this procedure on the gradient was either very small or else transient, lasting for less than 3 min.

For each experiment ten alternative chambers were used, one animal being introduced into each. One hour was allowed for the initial burst of activity to die down, after which the animals were transferred to the middle of the gradient by rotating the false floor. After a period the distribution of the animals was noted and they were moved back once more to the middle of the gradient. This process was continued until 130–250 position records were obtained. To assess the period of time taken for the complete establishment of the reaction, readings of distribution were taken initially every 3 min. This was continued until the ratio of animals in the wet to animals in the dry remained constant for five consecutive readings. This period was observed from five to eight times in each experiment and the mean value was found. Subsequently this period of time was allowed to elapse between moving the animals to the centre of the gradient and noting their positions.

It was not possible to run parallel control experiments, but such experiments, in which the humidity was the same on both sides of the chamber, were made at four different humidities. The results obtained did not differ significantly from those expected in the complete absence of any reaction.

Unfortunately Edney hygrometers were not available until almost all experiments were completed. At the end of each experiment the potassium hydroxide solutions were collected separately from the two sides of each chamber and the mean density for each side was determined. After the Edney hygrometers had been received, a series of chambers with hygrometers was set up and allowed to stand for 20 hr. At the end of this period the humidity in the two halves of each chamber was read on the hygrometers, the potash in each half was pooled and its density determined. From these observations it was possible to evaluate the humidity extremes above the false floor in any gradient. The values of the humidity gradients quoted below have been assessed in this manner. In uniform conditions the humidity has been directly determined from the density of the potash solutions.

The intensity of the reaction (I.R.) is expressed as the excess percentage of animals on the wet side, or 100 (W– D)/(W+D), where W is the number of animals in the wet and D the number in the dry. An animal lying in an area 1 in. (2·54 cm.) on either side of the midline was recorded as being neither in the wet nor the dry and was not included in the calculation of I.R.

A uniform humidity chamber was employed to determine the effects of humidity on various functions such as speed and frequency of locomotion. This is identical with the alternative chamber except that a uniform humidity prevails throughout the dish. The false floor was marked out in inch squares. This allowed the position of the animal to be recorded at any moment. From such position records the animals’ tracks could be drawn. The position of the posterior end of the animal was taken as representing that of the animal as a whole, for it was found that weaving movements of the anterior segments, introducing many small deviations, served only to complicate the tracking. The resultant path is more clearly represented by the tail track. Records were made every 15 sec. and the track was subsequently drawn on squared paper.

Speed of locomotion was determined by measuring the distance between successive time marks with an opisometer. The method of determining rate of turning has been described by Ullyott (1936). Deviations imposed upon the animal by ths circular nature of the apparatus have been disregarded. Since the speed was found to vary with humidity, it was undesirable to measure the rate of turning in degrees per minute. Instead it has been expressed as degrees per centimetre track. This unit will be referred to as ‘angular deviation’. Locomotory activity was determined by recording at 5 min. intervals over a period of 2 hr. Animals were classified as active when showing definite locomotory activity, inactive if there was a complete absence of movement and virtually inactive in all intermediate conditions.

An alternative air-current chamber was employed to establish the presence or absence of a klinokinetic reaction. The apparatus consisted of a uniform humidity chamber through which a steady current of air could be passed. The current was introduced through tubes which opened at four equally spaced points beneath the false floor; the outlet was in the roof of the chamber. The rate of flow of air was about 300 ml./min. The air current could be passed through concentrated or dilute potash solutions before entering the chamber. This provided currents of low and high humidity which could be alternated at will. In preliminary experiments the changes of humidity on reversing the air current were followed by means of a dew-point hygrometer whose cup was placed just above the false floor. The results showed that the change from one humidity to another was complete in less than 5 min. To observe whether there was effective replacement of one gas by another under such conditions, an air current charged with nitric oxide was passed into the chamber. No marked inequalities of concentration of nitrogen peroxide could be seen. The gas passed as a level cloud through the false floor and displacement was complete in 3–5 min. The conditions of these experiments conform fairly closely to those under which the adaptive element of klinokinesis was first demonstrated (Ullyott, 1936), in that the stimulus is non-directional and the change of level of stimulation, although by no means instantaneous, is fairly abrupt. Any adaptational phenomenon, unless very rapid, should be detectable.

The experimental procedure employed was as follows : Air, conditioned to a fairly high relative humidity (R.H.) was blown through the chamber for 2 hr. An animal was then introduced into the chamber and tracked for 25 min. The air current was then switched so that air of low R.H. passed through the chamber. Tracking was continued for a further 25 min. and then finally the current was switched back again to the original humidity and tracking continued for a final 25 min. period. The track records were divided into 10 min. intervals and the speed and angular deviation were determined for each period. Unfortunately, experiments starting with a lower humidity could not be undertaken owing to the adverse effect on the animals of prolonged exposure to dry conditions.

While this apparatus permits a fairly rapid change in ambient humidity without mechanical disturbance, it is open to the objection that the air current may modify the behaviour of the animal. In practice there appeared to be no significant difference between the values obtained for speed and angular deviation in the alternative current chamber and those observed in still air at the same humidities. The results in the two chambers are set out in Table 1.

Nevertheless, to test this objection a further series of observations was made using a double-humidity chamber.^* In this the false floor is supported by four peripheral and one central dish. Immediately above the latter a small watch-glass was inverted. The central dish and watch-glass were of the same diameter, so that they formed a small enclosed chamber. Attached to the watch-glass was a metal rod which projected through a small hole in the chamber roof. By means of this rod the watch-glass could be raised and suspended beneath the glass roof. Chambers were set up with the peripheral dishes containing dilute potash solution so as to give a high R.H. in the main chamber. The central dish contained a calculated excess of concentrated potash solution. The excess served to correct for any diffusion from the main chamber and thus to minimize humidity changes in the central chamber. The dishes were allowed to stand for 12 hr. before use. An animal was then confined in the central chamber. This operation necessitated removing the glass lid, but could be performed in a few seconds. The apparatus was then left for a further half an hour to condition the animal and allow the humidity conditions to be reestablished. After this period the watch-glass was raised and suspended beneath the roof and the movements of the animal were recorded. The experiment could be repeated using reversed humidity conditions between the chambers. The advantage of this type of apparatus in the study of adaptation in angular deviation is that the level of stimulation is changed almost instantaneously provided the animal is active, but we found that in this simple form it was unsatisfactory owing to the thigmotactic behaviour of Peripatopsis. It was not possible to raise the watch-glass without disturbing the animal mechanically, an event which it was particularly desired to avoid.

A divided alternative chamber was used to determine the nature of the directed reaction which was found. It consisted of an alternative chamber which had been divided into two halves by glass partitions above and below the false floor. Communication between the two halves of the chamber was afforded by four semicircular apertures of radius 2 mm. These were cut out of the lower edge of the upper plate and were equally spaced along its edge. A medium-sized Peripatopsis could pass fairly easily through such a hole.

Unless otherwise stated all experiments were carried out at 25·5 ± 0·3° C. A single neon lamp was used as a source of light. Preliminary experiments showed the absence of any reaction towards such a stimulation, but as a precautionary measure the alternative chambers were arranged symmetrically around the light source and rotated through 180⁰ half-way through each experiment.

When animals were activated in the chamber a period of about 15 min. usually elapsed before an equilibrium was reached. In many gradients the time taken for the development of the reaction to its full extent was observed. These results were examined to see whether the reaction developed more rapidly in some gradients than in others. No significant differences in time of development were found. The results for all humidity gradients have therefore been summed and the resulting curve is shown in Fig. 1. This is very similar to that found by Pielou & Gunn (1940) for Tenebrio molitor L., but equilibrium is reached slightly earlier in Peripatopsis. It differs markedly from the value of 1 hr. found by Bentley (1944) for Ptinus tectus Boie.

A study of the I.R. was undertaken to determine the preferred or eccritic humidity and to see how the I.R. varied with the range and absolute value of the humidity gradient. The reaction of an animal to humidity may be conditioned either by relative humidity as in Culex fatigans (Wiedemann) (Thomson, 1938) and Tenebrio molitor (Pielou & Gunn, 1940) or by saturation deficiency as in Agriotes larvae (Lees, 1943). Dakshinamurty (1948) finds Musca domestica L. to react to both. These possibilities may be distinguished by determining the I.R. over the same range of R.H. at different temperatures, since under these conditions the ranges of saturation deficit will be different. The results of such experiments are shown in Table 2. It will be seen that with a decrease in temperature and consequent fall in the range of saturation deficiency, the I.R. falls. In these experiments it was further observed that the time taken for the full development of the reaction was not affected by temperature. It seems therefore unlikely that the fall in I.R. with temperature is due simply to a decrease in activity. While the possibility of some other modifying effect of temperature cannot be definitely excluded, these results suggest that the reaction is governed by saturation deficiency rather than R.H. The difficulty of maintaining low constant temperatures precluded the extension of these observations.

Table 3 shows the intensity of reaction for various humidity gradients at 25·5° C. In Fig. 2 the I.R. is plotted against the highest available R.H. From these results it will be seen that there is a slight but definite reaction away from complete saturation, although over the greater part of the range the reaction is away from the drier conditions. There is thus at this temperature an eccritic humidity of about 98% resulting from the expression of these opposed tendencies. The absolute value of the I.R. is dependent on the range of the humidity gradient as well as on the higher R.H. It will further be noted that there is probably a complete absence of reaction at low humidities. Experiments in a very dry atmosphere could not be made owing to the high mortality which occurs under such conditions.

The structure of the dermal sense organs of Peripatopsis moseleyi has been described by Manton & Heatley (1937). Three types have been recognized. Thin hair-like sensilla occur on the antennae. Over the general surface of the body are further sensilla which differ from those on the antennae in having a comparatively thick cuticular covering. A third type is found on the lips and tongue. These differ from the previous types in projecting only slightly above the general surface of the skin and in being open at their tips.

To locate the position of humidity receptors it is necessary either to amputate the organ on which they are borne and show a subsequent change of behaviour (Pielou, 1940; Lees, 1943; Wigglesworth, 1941) or to render the organ functionless in some other way (Begg & Hogben, 1946). It was found possible to remove the antennae without otherwise damaging the animals ; specimens lacking their antennae survived under laboratory conditions for a period of over 6 months. For experimental purposes a recovery period of at least 48 hr. was allowed after the operation. Various methods were tried to render the sensilla of the body functionless. The only successful procedure was that of covering the body with vaseline. The vaseline was first smeared on a glass plate. This was warmed and then pressed against the region of the animal which was to be studied. A little fluorescein was mixed with the vaseline so that the exact extent of the area covered could be determined by examination in ultra-violet light. Only a portion of the body surface could be treated at any one time, for if the whole surface was covered the animal died rapidly. The buccal sense organs were also treated with vaseline applied on a fine glass rod which was inserted into the buccal cavity. Before this operation the animals were immobilized by cooling to o° C. for 10 min.

The effect of removing both antennae on the I.R. is shown in Table 4. It will be seen that over the range 95–78 % R.H. there is no significant difference between the I.R. of normal and operated animals. It is unlikely that this negative result is due to incomplete extirpation, as examination showed that the sensilla are most concentrated towards the tip of the antennae and that there are but few near its root. With a range of 99·8–98·2% R.H. the result is different. Here the reaction away from the wet is abolished, the I.R. of the antennectomized animals not differing significantly from the value obtained in control experiments where there was no gradient.

These results suggest that the receptors mediating the reaction away from complete saturation are not identical with those responsible for the reaction away from dry conditions. The former, but not the latter, would appear to be situated on the antennae. Confirmatory evidence for this suggestion will be presented later in a discussion of the mechanisms of the humidity reaction. The existence of an eccritic humidity of about 98% R.H. at 25·5° C. thus appears not to be due to the activity of a single type of receptor imposing a behaviour pattern by which the animal will aggregate at some humidity between saturation and dryness, but rather to be the resultant of two opposed reactions mediated by different receptor organs.

There is no evidence that the antennal humidity receptors are to be identified with the antennal spines.

All experiments were carried out in alternative chambers with a humidity gradient of about 94–75 % R.H. The controls were subjected to the same treatment as the experimental animals except that the glass plates were not vaselined. The results obtained by treating different areas are set out in Table 5. It is clear that treating the dorsal and lateral surfaces had no effect on the response. On the feet of Peripatopsis are heavy concentrations of sensory spines. These are mounted on pads on the median ventral surface of the appendages. Vaselining the feet and ventral surface can also be seen to be without effect on the l.R. It is apparent that attempts to identify the receptors by these methods were unsuccessful. However, experiments made with the divided alternative chamber indicated that humidity receptors are probably widely distributed over the body. These results will be discussed later.

A negative result was also obtained for the buccal sense organs.

The behaviour patterns which may bring about an aggregation of animals in response to diffuse stimuli have been considered by Fraenkel & Gunn (1940). They consider that there may exist four types of reaction which will produce such a response. We have approached the problem with these views in mind.

The simplest mechanism proposed is described as orthokinesis. This is an undirected reaction in which speed or frequency of locomotion is dependent on the intensity of stimulation. To determine whether this mechanism was present the speed and activity of animals were studied at various humidities in the uniform chamber. The effect of humidity on speed is shown in Fig. 3. It will be seen that speed is minimal at 98 % R.H., rising sharply on either side of this point, but levelling off at lower relative humidities. Humidity influences the activity of the animals as well as their speed of locomotion. Fig. 4 shows the effect. Activity and inactivity bear much the same relation to humidity as does speed, but virtual inactivity is independent of humidity within the limits of variability.

From Figs. 3 and 4 it is clear that the high level of speed and activity found at very high humidities will be partly or wholly responsible for the observed reaction away from saturation. It has previously been shown that removal of the antennae abolishes the reaction away from saturation in a gradient of 99·8–98·2% R.H. It is therefore to be expected that animals without antennae would neither move faster nor show greater activity at saturation than at 98% R.H. This expectation has been confirmed. Fig. 5 and Table 6 compare the behaviour of normal and antennectomized animals at different humidities. It will be seen that the activity of antennectomized animals at saturation is no greater than at 98% R.H., although the general level of activity for both normal and experimental animals is the same at lower humidities. A similar effect for speed of locomotion is shown in the table.

It thus appears that Peripatopsis displays a marked orthokinesis in both speed and activity. In mediating the reaction away from the dry this response is most effective between 85 and 98 % R.H. At the same time the experiments with antennectomized animals have shown that orthokinesis is also partly responsible for the reaction away from saturation, and thus it clearly plays an important part in giving an eccritic humidity of about 98% at 25·5°C.

A second method of aggregation in a non-directional gradient depends upon an increase in angular deviation with increased intensity of stimulation. Adaptation to the increased stimulation is an essential adjunct. This type of response has been demonstrated most clearly in the reaction of Dendrocoelum lacteum Oerst. to light (Ullyott, 1936).

In the study of this type of reaction the angular deviation was first determined at various humidities in the uniform chamber. The results are shown in Fig. 6. It will be seen that there is a significant decrease in the angular deviation with decreasing humidity. This result is different from that of Ullyott, where there was the same basal angular deviation at all intensities of illumination.

There was, however, some evidence of an adaptation in Peripatopsis, although the decrease in angular deviation with time was very irregular. It was thought that this might be an expression of some activation following the introduction of the animals into the chamber. To test the validity of this suggestion, experiments were carried out in the alternative air-current chamber so as to eliminate the effect of sudden mechanical disturbance. The results obtained are presented in Table 7 and Fig. 7. The experiments show that the angular deviation is maintained at a high level for 25 min. after the humidity of the air current is changed from 42 to 73 % R.H. There is no evidence of adaptation and, moreover, the basic angular deviation varies with stimulation intensity, an effect not found in true klinokinesis.

As has been stated above, these experiments are open to the objection that the pattern of behaviour may be modified by the air current. It is also possible that the rate of change of stimulation was so slow that adaptation might have proceeded as rapidly as the humidity changed. To check these results, experiments were carried out in the double-humidity chamber. The results obtained are set out in Table 8. Two control experiments were made in which both central and main chambers were adjusted to the same humidity. It will be seen that, while the angular deviation is reasonably constant at 80 % R.H., in the dry chamber the value falls gradually to a level not significantly different from that observed in the uniform humidity chamber (Fig. 6). This suggests that some element of mechanical disturbance is introduced in releasing the animals from the central chamber. The experimental results show the same pattern as the controls. This might be due to one or more of a number of causes; the important observation, however, is that where the animal moves from the dry to the wet the high level of angular deviation appears to be immediately established and to be maintained. The whole pattern is thus distinct from klino-kinesis with adaptation.

To make a rough assessment of the possible importance of these effects in contributing to an aggregation in the wet, a hypothetical track was drawn of an animal in an alternative chamber of which one-half was at 75 % R.H. and the other at 40 %. Appropriate values of mean angular deviation were used in each half of the chamber. The direction of turn was randomized by spinning a coin. The I.R. was calculated from the track lengths in the wet and in the dry and was found to be 5 %. It is thus clear that the turning movement, while slightly enhancing the wet preference, is not an important factor under these conditions. The effect will be increased in a more extreme gradient, but is unlikely to make a major contribution to the total reaction.

A formula devised by Thorpe, Crombie, Hill & Darrah (1947) makes it possible to assess approximately the intensity of a reaction mediated by orthokinesis when the relation between mean speed and humidity is known. When this formula is applied to the data obtained from Peripatopsis it appears that the reaction away from saturation can be completely accounted for by orthokinesis. Thus over the humidity range of 99·8–99·0% the expected value of the I.R. is 35·2%, which is more than adequate to account for the observed value of 19·7%. Over other parts of the gradient orthokinesis makes only a partial contribution to the reaction as a whole. Thus over a gradient of 39% R.H. with a higher relative humidity of 87%, the expected I.R. is 38·9%, while the observed value is 100%. Since the klinokinetic mechanism is clearly insufficient to account for this difference, a search was made for a directed reaction.

Two types of directed reaction are considered by Fraenkel & Gunn (1940) to be possible in a gradient. In the one, klinotaxis, successive intensity comparisons are made by regularly alternating deviations of a part or the whole of the body. The other, tropotaxis, is dependent on the simultaneous comparison of the intensity of stimulation of two separate receptors or groups of receptors on either side of the body. For the analysis of the directed reaction the divided alternative chamber was used. An animal was introduced into one-half of the chamber and its behaviour noted. It moved in a random fashion and would eventually reach one of the apertures leading to the other half. Here one of three things invariably occurred:

• (i) The animal went straight through the aperture and into the other half without hesitation.

• (ii) The animal moved partly through the aperture, locomotion ceased abruptly, and after a certain interval of time the animal backed into the half whence it came.

• (iii) As in (ii), the animal ceased to move for a time, but then, instead of backing out, it continued on its way into the other half of the chamber.

The behaviour described in (ii) is an ‘avoiding reaction’, while that shown in (i) and (iii) was considered as ‘no reaction’. Several different animals were used in a single experiment and each was observed for a period of 1–2 hr. The results are expressed as the percentage of the total number of trials which showed an avoiding reaction.

The first experiment was done with R.H. of o and 100% in the two halves of the chamber. The results are given in Table 9. It will be seen that the percentage avoidance for animals going from the wet to the dry is ten times that of animals moving in the opposite direction. In only one instance was an avoiding reaction to the wet observed, and this was made by an animal which had just been introduced into the chamber. The experiments were repeated using humidities of 80 and 50 % on the wet and dry sides respectively. In this case the percentage avoidance of the dry was 67 and of the wet 14. As a control the experiment was repeated with 50% R.H. on both sides of the barrier. The percentage avoidance was the same in both directions.

These results show that the directed reaction cannot be klinotactic in the strict sense of the word. Once the body of the animal is passing through the aperture the anterior regions of the body will be subjected to uniform stimulation in spite of any lateral deflexions which may be made, communication between the two halves of the chamber being blocked by the body of the animal. Lateral deflexion will not therefore afford an opportunity for the comparison of different intensities of stimulation.

The experiment does not exclude the possibility that the animal may achieve orientation by comparison of intensities which are successive in time, but independent of any lateral movement. Such a klinotaxis has been shown by Lees (1943) in the larvae of Agriotes. The only other manner in which orientation could be carried out would be by a comparison of the intensities stimulating the anterior and posterior parts of the body. In other words by a tropotaxis based on a comparison of stimulation of anterior and posterior receptors rather than of bilaterally symmetrical ones.

To distinguish between these alternatives, experiments were carried out in which the posterior end of the animal was smeared with vaseline. This treatment was assumed to prevent evaporation and thus be to the animal equivalent to stimulation by a saturated atmosphere. The treated animals were introduced into the drier half of a divided alternative chamber in which the R.H. was 50%; the R.H. on the other side was 80%. If the directed reaction depended upon successive comparisons in time it would be expected that under these conditions the animals would preferentially pass into the moister half of the chamber. If, however, the reaction depended upon a comparison of anterior and posterior ends, it would be expected that the animal would back into the drier half. In the experiment it was found that, while the percentage avoidance of the dry by animals which had passed into the wet half of the chamber was in good agreement with the controls (χ^{2 =} 0·54; P = 0·5), the percentage avoidance of the wet by animals moving out of the dry was 67 compared with a value of 14 for the controls (χ² = 8·78 ; P = 0·01). In other words, two out of every three treated animals backed into the dry while only three out of fourteen normal animals did so. An avoiding reaction was therefore performed in spite of the fact that it had the effect of bringing the animal back into the drier half of the chamber. These results are most easily explained on the assumption that the directed reaction of Peripatopsis is based on a comparison of humidities at the anterior and posterior ends of the body.

This hypothesis is supported by the observation that animals whose posterior ends have been smeared with vaseline turn abruptly through angles of 180° more frequently than do untreated controls. The average number of such turns made by treated animals was 5·5 + 0·6 in 15 min., while the corresponding value for untreated controls was 2·5 ±0·3. Both sets of observations were made at 80% R.H. This phenomenon can probably be regarded as a novel expression of the traditional ‘circus movements’.

It should be remarked that the smearing of the posterior regions of the animals caused their death within 12 hr., and that within 3 hr. of smearing the treated half of the body was rendered functionless as regards locomotion. For this reason no attempt was made to confirm these observations using an alternative chamber. In using the divided chamber the animals were allowed half an hour to recover from treatment and then observed for a period of less than 2 hr. It therefore seems unlikely that these considerations need invalidate the results.

It should also be emphasized that these experiments do not exclude the possibility that orthodox klinotaxis exists. They show that it cannot be the basis of the avoiding reaction, but the possibility remains that lateral deviations of the anterior end of the organism assist in orientation. We have been unable to devise experiments which would enable us to test this possibility.

It may be concluded that the pattern of humidity behaviour of Peripatopsis includes a directed reaction which is dependent upon a simultaneous comparison of the stimulation intensity at the two ends of the body. In addition, a complex orthokinetic reaction is responsible both for a reinforcement of the directed reaction away from the dry and for an avoidance of complete saturation. There thus results a preferred humidity of about 98% R.H. at 25·5° C.

A few experiments have been made to study the effect of desiccation upon the intensity of the reaction. These are summarized in Table 10. It will be seen that there is a significant reduction of the I.R. with desiccation.

This effect might be due to two possible causes. One is that desiccation is accompanied by a relative increase in the speed and level of activity at the higher humidities compared with the lower. This would result in a weakening of the orthokinetic contribution to the reaction and the I.R. would thus fall. The other possibility arises from the character of the humidity receptors. Evidence has been brought forward to show that these depend upon the rate of evaporation of water. It has been shown by Manton & Heatley (1937) that the rate of evaporation of water falls off with desiccation in Peripatopsis. It is therefore possible that the sensitivity of the humidity receptors falls and the tropotactic reaction is weakened. The orthokinetic reaction may also be less intense as a result of lowered speed and activity at lower humidities. Experiments to distinguish between these possibilities have yet to be undertaken.

The first point of interest in the behaviour of Peripatopsis is clearly the marked preference for a humidity of about 98% R.H. at 25·5° C. In the case of most arthropods which have been studied, a more or less clearly marked preference, either for saturation or for dryness, has been discovered. Sometimes this preference is expressed over the whole humidity range. Thus Pielou & Gunn (1940) have shown a weak but definite preference for dryness in Tenebrio molitor in a gradient of 5–10% R.H., while Lees (1943) has shown a preference for saturation in Agriotes larvae over a range of 99·5–100% R.H. Frequently the preference is not strongly marked outside a certain range of humidity. Thus Wigglesworth (1941) has shown that Pediculus humanis corporis de Geer avoids humidities of 95 % and above, but is indifferent to lower values. Similarly, Gunn (1937) has found Porcellio scaber (Latreille) to prefer humid conditions but to be indifferent in gradients within the range of 80–100% R.H.

A preferred humidity of 60–80% R.H. has been shown to exist for blood-fed females of Culex fatigans by Thomson (1938), while Begg & Hogben (1946) claim that Drosophila melanogaster Meigen has a preferred humidity of 90 % ; their published figures do not, however, fully support this idea. We have shown that in Peripatopsis the preference is the result of the opposing action of two sets of receptors. Begg & Hogben have also demonstrated the presence of two sets of receptors by their experiments on mutant Drosophila. They have shown the presence of antennal receptors which mediate a reaction towards the wet and of other unidentified receptors mediating a reaction towards the dry. They point out that there is the possibility that the two opposed behaviour patterns may be dependent upon the stimulation of identical types of receptors which have different nervous connexions. This possibility cannot be excluded in Peripatopsis.

The existence of a clearly marked eccritic humidity recalls the existence of temperature preferences in many animals. It was felt that it would be of interest to see how far the two cases are parallel. Unfortunately, the methods by which preferred temperatures are attained have been little studied. Analysing earlier work Fraenkel & Gunn (1940) consider that klinokinetic reactions away from both extremes play an important part. Wigglesworth (1941) has found this to be so in Pediculus, but in two other recently investigated cases (Agriotes, Falconer, 1945 ; Ptinus, Gunn & Walshe, 1942) this possibility seems to have been excluded, and aggregation is attributed in both instances to a ‘shock reaction’ which has not been further analysed. These results find a parallel in Culex, where Thomson (1938) has described a ‘hygrophobotactic’ response away from both extremes of humidity. The condition in Peripatopsis, where the avoidance of one extreme is dependent on orthokinesis alone, has no parallel in investigated examples of temperature preferences.

The biological significance of this eccritic humidity for Peripatopsis may lie in the susceptibility of this animal to fungal infection. It is likely that by avoiding conditions of complete saturation, the animal is assisted in avoiding this hazard.

A second feature of interest is the decline in the I.R. with desiccation. Where this effect has previously been examined in animals showing a dry preference, that preference has been weakened as in Locusta migratoria migratorioides R. & F. (Kennedy, 1937) and Culex (Thomson, 1938), or the reaction has been reversed as in Blatta orientalis L. (Gunn & Cosway, 1938), Ptinus (Bentley, 1944) and Ixodes ricinus L. (Lees, 1948). Cases are also known where desiccation initiates a reaction towards the wet where previously none could be detected. In only one case have details of the effect of desiccation on an animal showing a wet preference been published. This is in Drosophila (Begg & Hogben, 1946), where the wet preference is strengthened. It will be seen that all these cases differ from that of Peripatopsis in so far as the behaviour following desiccation assists these other animals to a greater or less degree to reduce water loss. In its unteleological behaviour Peripatopsis is at the moment unique. It will require further investigation to see whether this effect is indeed a concomitant of the evaporimeter type of receptor, as has been suggested above, and whether compensating nervous mechanisms have been developed in other forms.

1. The behaviour of Peripatopsis moseleyi (Wood-Mason) towards humidity has been investigated. At 25·5° C. the animal is found to have a preferred relative humidity of about 98 %.

2. Humidity receptors mediating a reaction towards the wet occur over the general surface of the body. Receptors mediating a reaction towards the dry occur on the antennae. The receptors on the body surface probably depend for their functioning upon the rate of evaporation of water.

3. The reaction towards the wet is effected by an orthokinesis involving both speed and activity. This is enhanced by a tendency to turn more frequently in the wet than in the dry. There is further a directed reaction which depends upon the simultaneous comparison of the stimulation intensities at the anterior and posterior ends of the animal.

4. The reaction away from saturation is due to a speed and activity orthokinesis.

5. The existence of a reaction away from saturation together with the opposing reactions away from dryness result in a preferred humidity just below saturation.

Our thanks are due to Dr R. F. Lawrence of the Natal Museum for generously placing at our disposal his extensive knowledge of the biology of Peripatopsis. During the course of this work one of us (E.B.) held a bursary of the South African Council for Scientific and Industrial Research. We are grateful to the Council for a grant towards the expense of this investigation.