(1) The flame cell of platyhelminths is a composite organ formed from two cells. One cell contains a large nucleus and bears the flagella which form the flame. The other cell which contributes the barrel is the first tubule cell. There is a region of interdigitation between the two cells at the top of the barrel and the interdigitations are joined along their length by desmosomes. The tubule lumina are extracellular, the smaller tubules, at least, being formed by encircling projections from a single cell, joined at their tips by desmosomes. Cestodes apparently have no desmosomes in their tubule walls so that the tubule lumina may be intracellular. The tubules of most platyhelminths are lined by folds or microvilli, and flagella may be present in the lumen. The protonephridia of nemertines, entoprocts and priapulids appear to be of this type.

(2) Direct evidence of the physiological role of flame cell systems is limited. There is, for example, no proven instance of the production of urine hypo‐osmotic to body fluids by any fresh‐water platyhelminth or nemertine. Endoparasitic platyhelminths are apparently unable to osmoregulate. The relative permeability of the body surface of marine, fresh‐water, terrestrial and parasitic platyhelminths and nemertines may be related to protonephridial function.

It seems highly likely that the function of the flame cell is to filter interstitial fluid, separating water and crystalloids from macromolecules. The ultrafiltrate produced then flows down the tubules as a result of the hydrostatic pressure generated by the beating of flame flagella, or as a result of peristaltic waves of the whole body generated by the musculature of the worm. The fluid may be modified in the canal lumen by both active and passive resorption of solutes or the secretion of material from the walls into the lumen. Experiments with the larger platyhelminths suggest that the main function of the system is to remove organic metabolites from the interstitial spaces of the deeper tissues of the worm by a mechanism more efficient than simple diffusion.

(3) The flame bulbs of rotifers are fan‐shaped and the nucleus is in the tubule not the cap. The barrel of the flame bulb is composed of a series of columns in scalloped formation, each arc of the scallop being supported by a cytoplasmic pillar. A membrane interconnects the columns and each column is linked to its neighbouring central pillar by fibrils. The tubules leading from the flame bulbs are a complex system of three to four multinucleate cells. They empty into a contractile bladder. The protonephridia of acanthocephalans and gastrotrichs may be of this type.

(4) The mode of action of flame bulbs is probably to filter the pseudocoelomic fluid which is then modified by selective reabsorption in the tubule system. Rotifers are able to osmoregulate and this may be the chief function of their protonephridia.

(5) Solenocytes are morphologically diverse, usually with a cytoplasmic cap containing a nucleus, and a long tubule, in the lumen of which lie one or two flagella. The walls of the tubule are pierced by fenestrations, probably the site of fluid ‘filtration’.

(6) Kümmel (1962) has suggested that the many types of terminal organ are the result of divergent evolution from a single ancestral type. We suggest that protone‐phridial terminal organs can be divided on structural grounds into three or four different groups which are probably not inter‐related. This would mean that the apparent structural similarities which do appear would be the result of convergent evohtion imposing a conformity based on functional requirements.