Canal systems of the third type with tubular chambers are an improvement on those of the second, in that the area of choanocytes is increased by the pouching of the chamber-layer without corresponding increase in the size of the sponge. However, the area of choanocytes represents expenditure of energy, and the next problem to be solved is how to retain the improved current and at the same time to cut down expense. The first step is to change the form of the chamber from tubular to spherical. Now the energy of all the choanocytes is concentrated on the same small volume of water. The area of choanocytes is less, but the end result is as good as before. At the wide mouth of the spherical chamber there is nevertheless still a cause of loss of energy in the form of eddies, and it is as an obviation of these that one must regard the aphodi and prosodi with which higher members of the Demospongiae are provided. The correctness of this view receives support, apart from mechanical principles, from the fact that the mass of the body of any one of these sponges is greater relatively to the total flagellated area than in those sponges with eurypylous chambers; that is to say, a few {237}aphodal and diplodal chambers are as efficient as many of the eurypylous type. It is manifest that the current is the bearer of the supply of FOOD; but it requires more care to discover (1) what is the nature of the food; (2) by which of the cells bathed by the current the food is captured and by which digested. The answer to the latter question has long been sought by experimenters,[275] who supplied the living sponge with finely powdered coloured matters, such as carmine, indigo, charcoal, suspended in water. The results received conflicting interpretations until it became recognised that it was essential to take into account the length of time during which the sponge had been fed before its tissues were subjected to microscopic examination. Vosmaer and Pekelharing obtained the following facts: _Spongilla lacustris_ and _Sycon ciliatum_, when killed after feeding for from half an hour to two hours with milk or carmine, contain these substances in abundance in the bodies of the choanocytes and to a slight degree in the deeper cells of the dermal tissue; after feeding for twenty-four hours the proportions are reversed, and if a period of existence in water uncharged with carmine intervenes between the long feed and death then the chambers are completely free from carmine. These are perhaps the most conclusive experiments yet described, and they show that the choanocytes ingest solid particles and that the amoeboid cells of the dermal layer receive the ingested matter from them. In all probability it is fair to argue from these facts that solid particles of matter suitable to form food for the sponge are similarly dealt with by it and undergo digestion in the dermal cells.