W.H. Thorpe, Learning and instinct in animals, Methuen, Londres, 1956.

Perhaps an even more remarkable example of such flexibility in repair is shown by the work of Dembowski (1933) on the tube-building of the caddis larva of the genus Molanna (Fig. 5), quoted by Russell (1945). If deprived of its case, the insect will build a new one of the same pattern, and this it does by burying itself in the sand, backing downwards, and pushing a mound of sand grains towards its head, at the same time binding them into a loose bundle with silk from the labial glands.  This form an anchor to which the insect then attaches its anal hooks. Having done this, it makes a ring or section of tube round its body, carefully selecting suitable pieces to fit into the growing mosaic by adding material from the anterior edge  from the holdfast. Wings are built by stretching at full length out of the tube, but as the insect  always keeps its anal hooks within the tube, it cannot reach back far enough to complete the hinder part of the wings, and it usually achieves the normal and typical shape by lengthening the tube in front and cutting it off behind. Cutting away an unwanted part may also occur as follows: if a number of larvae are ejected from their cases but allowed to return they fail to get back to theright ones, and some of course get cases which are too large and others too small. The larva which gets into too large a case first cuts off a piece of the front of the tube and then a piece of the overhanging roof, as shown in the accompanying illustration (ig. 5), thus getting once more a tube of the right size. But the power of adaptation to accidental injury of the case does not ed here. If the roof is damaged, the missing part may be simply replaced. If the hinder part or up to 15 per cent. of the case is removed, an extra piece will be built on in front and the posterior part of the case is cut away. The insect may then extend it in front as before, or alternatively it may turn round and restore the missing part and then turn back -then the variety of response is very great. Dembowski described six main ones. Firstly, a new case may be built, a pieces of the old one merely being uses as a holdfast which is later cut away. Secondly, the case may be repaired by building on its front. Thirdly, repair may be accomplished by building on at the back and then turning and cutting off the roof and front part of the tube, then rebuilding the former. Fourthly, there is the possibility of building a new front end on to the hinder part and inhabiting it, but leaving the original front intact so that we now have a tube with two fronts. Fifthly, a new front end may be built and the rest trimmed off so as to create a permanently reversed tubes. Lastly, the insect may buid first in front and then cut off the new roof, reverse and build the hind end. In all these examples we get the impression that the goal is not a sequence of actions, or not that alone, but the construction of a case, or nest, of a particular pattern. It is true that these observations are in one sense unsatisfactory, in that the experiments should really have been performed only with animals which were constructing their cases for the first time. For otherwise it is always possible to argue that the previous eexperience of building a completed nest or tube by an innate chain-reflex series of actions may have resulted in a Learning of the details of the goal object. Indeed, Dembowski has shown that there is improvement in case-building with repetition of experience, for when four consecutive cases are built by the same individual, the fourth is better in a number of respects that is the first. But, whether previous expériences does not lead to the Learning of détails of thefinal stimulus, this work does very strongly suggest the exiostence of an extremely short-term purpose guiding the actions towards a goal. Further observations on the process by which Molanna selects the material for its case are provided by Szlep (1958), Frankhauser and Reik (1935) in America have published accounts of the nest-building of the caddis-fly larva, Neuronia postica, when supplied with abnormal materials for making the cases and when the first and second pairs of larval legs are lacking. Even in these conditions, the larva showed remarkable adaptability in achieving the construction of cases which were surprisingly near normality. Dudziak (1950) has carried out a similar study on the Caddis Phryganea and Miklaszewska (1948) on the aquatic caterpillar of Nymphula Brickenstein (1955) finds however that the snare construction of the caddis Neureclipsis is very dependent on water-current and other conditions.