Journal of General Microbiology (1991), 137, 735-743. Printed in Great Britain 735 The permittistat: a novel type of turbidostat GERARD H. MARKX, CHRISTOPHER L. DAVEY and DOUGLAS B. KELL* Department of Biological Sciences, University College of Wales, Aberystwyth, Dyfed S Y23 3DA, UK (Received 21 August 1990; revised 22 November 1990; accepted 7 January 1991) Baker’s yeast was grown in a novel type of turbidostat in which the steady-state biomass level was controlled not by the optical turbidity but by the dielectric permittivity of the suspension at appropriate radio frequencies. Dry weight, fresh weight, the optical density at 600 nm, percentage viability (from methylene blue staining), bud count and ethanol concentration were measured off-line and the cell size distribution was recorded using flow cytometry. Any changes in the physiological properties of the yeast had a negligible effect on the ratio between the permittivity set (and measured) and the steady-state dry weight, fresh weight or optical density of the cultures. The permittistat was found to provide an extremely convenient means for carrying out turbidostatic culture. Introduction Amongst the possible forms of continuous culture (Tempest, 1970; Pirt 1975), it is usual to distinguish those (such as the chemostat) in which the dilution rate is fixed by the experimenter and growth is nutrient-limited, and those, such as the turbidostat (Bryson & Szybalski, 1952) or pHauxostat (Martin & Hempfling, 1976) in which cell growth is not nutrient-limited and in which the inflow of fresh nutrients is controlled via a feedback loop responding to some growth-associated parameter. Al- though the chemostatic and turbidostatic modes of growth are fundamentally similar (Herbert, 1959, the volumetric productivity of biomass displayed by turbido- stats is the greatest achievable for a given organism and environment, and, from a control engineering point of view, the system is stable at high dilution rates, corresponding to values approaching pmax. In a turbidostat the biomass level in the fermenter is determined continuously, the nutrient pump being turned on when it exceeds a set point. The usual method of biomass estimation is via the optical turbidity of the cell suspension (Myers & Clark, 1944; Bryson & Szybalski, 1952; Anderson, 1953, 1956; Northrop, 1954; Moss, 1956; Cooper et al., 1959; Schlecht et al., 1958; Herbert et al., 1965; Dean, 1967; Moss & Bush, 1967; Blachere & Jamart, 1969; Munson, 1970; Edwards et al., 1972; Agrawal, 1987). However, turbidity measurements have two major drawbacks: (i) they are only linear (in * Author for correspondence. email : DBK@UK.AC. ABERYSTWYTH. terms of the Beer-Lambert law) over a limited range of values of biomass, and become useless at optical densities exceeding approximately 4 (corresponding for a 1 cm path length to approximately 2 mg dry wt ml-l), and (ii) the optical sensors used in turbidity measure- ments are extremely prone to fouling by the micro- organisms which they are attempting to measure. In addition, non-cellular particulate material in suspension will contribute to the signal. An important, and now perhaps more popular, alternative to the optical turbidostat is the pHauxostat (Martin & Hempfling, 1976; Stouthamer & Bettenhaus- sen, 1976; Oltmann et al., 1978; MacBean et al., 1979; Bungay et al., 1981 ; Rice & Hempfling, 1985; Minkevich et al., 1989; Fraleigh et al., 1989, 1990; von Schulthess et al., 1990), in which the growth-associated microbial production of acid (or, in principle, base) causes a change in pH which is returned to its set-point not by the addition of alkali per se but by the addition of a more alkaline nutrient medium; the biomass level in the steady state is then determined by the buffering power of the medium (whilst the dilution rate again corresponds to a value approaching pmax for the medium and conditions employed). The pHauxostat is relatively straightforward to implement, but has the disadvantages that (i) the biomass level is still set indirectly, (ii) there is a limit to the range of buffering powers which can be provided, and (iii) the cells must actually change the external pH by a substantial amount as a result of their catabolic activities [which is not always the case (Watson, 1972; Firstenberg-Eden & Eden, 1984)l. Other indirect meth- ods for the estimation of biomass include the electrical 0001-6457 O 1991 SGM