Pergamon Progressin Neurobiology Vol. 44, pp. 233 to 247, 1994 Copyright © 1994Elsevier ScienceLtd Printed in Great Britain. All rights reserved 0301-0082/94/$26.00 ARE YOU USING NEURONAL DENSITIES, SYNAPTIC DENSITIES OR NEUROCHEMICAL DENSITIES AS YOUR DEFINITIVE DATA? THERE IS A BETTER WAY TO GO D. E. OORSCHOT Department of Anatomy and Structural Biology, and the Neuroscience Research Centre, University of Otago, P.O. Box 913, Dunedin, New Zealand CONTENTS 1. Introduction 2. Why should density measures be avoided? 2.1. Illustration of the importance of knowing the total reference volume 2.2. The effect of tissue processing on the number in a subvolume (Nv) 2.3. Volume changes and ratios derived from density measures 3. Calculating absolute or total numbers 3.1. Estimation of the absolute or total reference volume 3.2. Estimating the Nv or numerical density 3.2.1. The disector 3.2.2. The optical disector 3.2.3. Remembering potential processing effects on the final Nv 3.2.4. The fractionator 3.2.5. The "old and biased" versus the "new and unbiased" methods 4. Calculating absolute or total neurochemical amounts 5. The way ahead Acknowledgements References 233 233 234 236 237 238 238 239 239 241 242 242 243 244 244 244 244 1. INTRODUCTION Within neurobiology, a substantial number of published papers demonstrate an unfortunate meth- odological problem, namely that density measure- ments (i.e. the number of neurons or synapses per mm 2 or mm 3 of tissue, or the levels of a chemical in #g/rag of tissue or of protein) are used s the definitive or decisive data. Density measures, however, are parameters to be avoided because they can lead to serious misinterpretations of biological processes (as illustrated by Lange et al., 1976; Wagner et al., 1983; Bedi, 1984; Braendgaard and Gundersen, 1986; Haug, 1986; Swaab and Uylings, 1987; Pakkenberg and Gundersen, 1988; West and Gundersen, 1990; Bolender et al., 1991; Oorschot et al., 1991; Gundersen, 1992). Unfortunately, all of these articles have either been published in specialised books or journals, and/or in the context of a particular experimental field or technique. The result is that the important principles within these articles have largely been missed, and consequently ignored by the wider neurobiological community. Evidence for this is found in recent editions of three leading neurobiological journals (namely Brain Research 591; the Journal of Comparative Neurology 323; and the Journal of Neuroscience 12(7)). Of the papers reporting in vivo neuronal numbers, synaptic numbers or neurochemi- cal measurements in these three volumes, 15 out of 17 papers (88%) used density measures as definitive data. Moreover, it is not a trend peculiar to these particular volumes. Earlier volumes of Brain Research yielded similar results: in volume 570, 88% (i.e. 15 out of 17 papers) used density measures as definitive data; in volume 549, 92% (i.e. 11 out of 12 papers) did (with these earlier volumes being chosen systematically, after a random start, using the principles to be outlined in Section 3). Thus, the aim of the present review is to explain to the wider neurobiological community why it is that the use of density measures as decisive data is problematic and should be avoided. Thereafter, better methods will be suggested in the hope that this problem will be overcome. These suggestions will include a timely critique of: (i) the "old and biased" versus the "new and unbiased" methods, and (ii) the use of neurochemical density data as definitive data. Both of these areas have not been addressed in previous reviews, yet they are areas of considerable importance. This is especially the case for neurochemi- cal density data, since it is an area of equivalent concern to morphometric data in its potential to yield results and conclusions of limited scientific value. 2. WHY SHOULD DENSITY MEASURES BE AVOIDED? Density measures give an indication of the number of neurons and synapses, or the concentrations of 233