PHYSIOLOGICAL SOCIETY, JUNE 1981 is brought into focus. With video-image analysis and stage-stepping motors under computer control it is possible to focus automatically. Coleman, Garvey, Young & Simon (1977) developed an auto-focus for Golgi material which selected the position of darkest image. However, we have found it more appropriate to use the position of maximum smoothed-edge contrast for less-densely or less-uniformly stained neurones, e.g. those injected with Procion Yellow or HRP. The auto-focus described here was designed to allow the operator to concentrate fully on the location and classification of structures in the (XY) image plane and on error correction. As a bonus, it was found to be faster and more reliable than manual focusing.The Magiscan (Joyce Loebl) was used to analyse the video image automatically and to control the microscope stage. A high-power ( x 50) objective was used to sharpen the focus plane. Each neuronal structure was traced in the XY plane with a sequence of point pairs, either by automatic edge-tracing or manually with a light pen. A four pixel (approximately 0 7 jsm) square was centred on each of these two reference points. Grey levels from the sixteen outermost points of each square were used to estimate smoothed-edge contrast; the degree of focus was defined as the sum of the two maximum values obtained when adding grey levels from each four successive points along the edges of the two squares and subtracting those from the next four points. The video image was analysed at sixty-four levels of grey. At this sensitivity the noise in our standard Bosch camera resulted in fluctuations in grey readings of up to + 2 levels. The resulting variability in Z estimates was reduced to within ± 1 ,um step by adding results from two (25 Hz) video scans and smoothing the degree-of-focus curve (adding results from each three successive focus positions). The stability and sensitivity of this final measure was such that a fast adaptive scanning procedure could be used, mimicking manual focusing; instead oftesting over a relatively wide fixed range, the upward and downward movements were each stopped as soon as the focus measure was less than 98 % ofits maximum value (usually within 3 ,um steps).Supported by the M.R.C. REFERENCE COLEMAN, P. D., GARVEY, C. F., YOUNG, J. H. & SIMON, W. (1977 The results for the different anaesthetics showed trends in which the hydrocarbons lay at one extreme and the alcohols at the other. At concentrations which suppressed the Na current (at the potential ofits peak under control conditions) by approximately 50 %, the hydrocarbons moved the mid-point of the steady state inactivation (hz) curve and the peak of the inactivation time constant (Th) by ca. 15 mV in the hyperpolarizing direction. n-Pentanol, n-octanol and n-C8H17(0. CH2CH2)30H, by contrast, produced shifts (in the same direction) of less than 1 to 2 mV. The effects of the hydrocarbons and alcohols on the mid-point of the steady state activation (moo) curve on the peak of the activation time ...