The critical micelle concentration (CMC) of the principal lung surfactant phospholipid, L-a-dipalmitoyl phosphatidyl-choline, in water has the extremely low value of 4.60.5 x 1010M (ref. 2). In consequence the overwhelmingly preferred domain for a single phospholipid molecule in water is in association with other phospholipid molecules. The resulting aggregates, technically referred to as smectic mesophases or liposomes, come to occupy a minimum space in the hostile environment of water. The reason for the low CMC of the molecule is the large and incompatible-in-water hydrocarbon moiety. A liposome infrastructure3 is characterised by alternating bimolecular sheets of lipid intercalated by aqueous spaces, whereby each and every lipid sheet forms a 'closed' membrane system and where all the hydrocarbon chains are screened from water by the polar head-group region (Fig. \a). It seems unlikely that this material, in a fully hydrated state, could facilitate a rapid extension of an air/fluid interface by donating surface-active molecules, for example, as occurs during the first in-spiratory breath of an infant. Nevertheless there has been a tacit assumption that the lamellar bodies of alveolar type 2 cells are packages of lung surfactant, smectic mesophase in character, in equilibrium with cell water.We suggest that the lamellar bodies are packages of phospholipids in equilibrium with restricted amounts of water-less than about 15 water molecules per phospholipid, at which proportion the bimolecular sheets of phospholipid are not closed but 'open' and the structure has more of the properties of a water-in-oil system than an oil-in-water system. On release from a cell close to an advancing air/fluid interface the phospholipids would rapidly and easily donate molecules to the interface, since these molecules do not have to surmount the unfavourable hydration barrier (Fig. \b). The surfactant used in our experiments was prepared by washing out sheep's lungs with saline, removing the cellular debris by centrifugation, concentrating the fluid by lyophilisa-tion and extracting the complete lipid content by the Folch procedure4. The chloroform layer was evaporated to dryness under nitrogen leaving a dry, waxy material, the surfactant. This has been analysed and found to contain all the generally accepted surfactant lipids in normal proportions (J. Harwood, personal communication). In the experiments using wetted surfactant, liposomes were formed by sonicating the dry surfactant in saline. Surface tension measurements were made using a roughened platinum dipping plate (2 cm wide) suspended from a force transducer feeding into a recorder. The liquid was contained in a clean, Teflon trough.When a particle of dry surfactant was placed on the clean surface of physiological saline the surface tension always fell rapidly to the equilibrium tension of 24 dyn cm"1 (Fig. 2a). If a similar amount of wet surfactant was added to the surface there was either no effect on surface tension or else it fell very slowly to some variable, intermediate value (Fig. 2b). In other experiments we rapidly removed as much of a surface monolayer as possible by aspiration with a fine tipped sucker (Fig. 2 at each arrow). When the surface layer had been formed from a particle of dry surfactant the surface tension rose with aspiration and then quickly fell again, within 20 s, showing that more molecules were being donated to the surface. This could be repeated many times, but only as long as there was a particle of surfactant at the surface to act as a molecular reservoir (Fig. 2a). Once the particle was removed further recruitment of molecules to the surface became very slow. When wet surfactant was used and the surface monolayer was aspirated, the surface tension rose and fell again very slowly because few molecules were being recruited to the surface (Fig. 2b).The effect of dry surfactant on lung expansion was investigated by placing a small particle into the fluid-filled trachea of dead 27-d foetal rabbits (term is 31 d, and the lungs contain little endogenous surfactant at this age). Twenty-one foetuses from six litters were used; 22 alternate foetuses were used as untreated controls. A tracheostomy tube was connected to a closed circuit apparatus constructed to inflate and deflate the lungs and to plot pressure volume curves. Three cycles were performed on each lung at 37 C. Twenty-nine foetuses were subjected to pressures up to 35 cm H2O and 14 foetuses to pressures up to 30cm H2O. There was no difference in body weights and wet lung weights between the treated and control foetuses. All the 21 treated lungs expanded at opening pressures below 30 cm H2O. Only 6 of the 22 untreated lungs expanded, all with opening pressures above 30 cm H2O. The mean volume retained at atmospheric pressure was 0.410.75 ml (s.e.m.) for the treated lungs and 0.0380.012 ml for the untreated. These differences are statistically significant. Fig. 1 The molecular configuration of dry and wet surfactant. b, The dry surfactant has open ended layers from which molecules can spread freely to an air/water interface, a, In water the surf-factant aggregates as a smectic mesophase and few molecules are free to reach the interface.Fig. 2 The surface tension lowering properties of a, dry surfactant and b, wet surfactant placed on saline. In both instances surfactant was placed on the surface at 'On'. The dry particle was removed from the surface at 'Off'. Upwards-pointing arrows indicate when the surface monolayer was partly removed by aspiration. An intact surfactant monolayer is vital to the healthy functioning of the lungs. It allows the lungs to expand with a minimal expenditure of energy and prevents atelectasis in expiration. To be effective a surfactant must maintain the monolayer from the moment of birth. If the surfactant in the lungs is fully hydrated it will not form a surface active layer easily or quickly and will not spread fast enough to maintain an intact surface film. From these experiments we have shown that the active form of surfactant is the 'dry' state. We postulate that the lamellar bodies in the alveolar type 2 cells, which are known to be surfactant stored ready for release into the alveoli, are packages of surfactant in a 'dry' state. There have been many attempts to treat respiratory distress syndrome in the newborn with surfactant substances as a mist nebulised with water. Not surprisingly this has proved uniformly ineffective. Our results suggest that it may be possible to correct surfactant deficient states with 4dry' surfactant.This work was supported by the Nuffield Foundation, the MRC and the National Fund for Research into Crippling Diseases. C.J.M. thanks Professor G.S. Dawes for accommodation, facilities and help. We also thank D. J. Klass for advice.