We appreciate this opportunity to clarify our position on a few central issues of this debate. By our rejection of a number of Palaeozoic palaeomagnetic data from Scotland (Storetvedt et al. 1990), Torsvik et al. accuse us of disregarding all classical field tests. We have never disregarded any classical field test, but would argue that such tests have sometimes been misapplied. Also, we obviously have not rejected all earlier British 'SiluroDevonian' data, including Scottish results (e.g. table 3, Storetvedt et al. 1990).In our discussion of the data of Torsvik ef al. (1989), [see Storetvedt et al. 1990, fig. 10(b); Torsvik et al. 1991, this volume, fig. 4(b)], our objective was to point out the dangers of 'incomplete remanence analyses' in palaeomagnetism. Torsvik et al. now provide us with the order of demagnetization steps in their original data. With the information we had in 1989, we qualitatively suggested an alternate remanence model that fits the data. We wish to emphasize the following points. (1) We did not suggest at any time that T is representative of a real palaeomagnetic direction as attributed to us by Torsvik et al. (1991, this volume). (2) We recognized T as a trajectory on the vertical plot only with no corresponding branch in the horizontal plot (Storetvedt et al. 1990, p. 160), and we argued that this is probably the result of partial spectra overlap of two components. With the new information on temperature steps and their sequential order (Torsvik et al. 1991, this volume), we consider our original point as well emphasized. A perusal of fig. 4(a) (broken lines) of Torsvik et al. (1991, this volume) indicates that their interpretation of this sample's data is not sustainable: in computing the high-temperature component they admit that the line above 570 "C was origin-anchored, a procedure which forces the data into a pre-conceived frame. To support their original interpretation of sample L283 Torsvik et al. now submit results from two additional specimens ( fig. 2), but in cases of complex magnetization it is often possible to find individual examples to bolster any pre-conceived view. What really counts is a holistic understanding of the total database (aided by vector analysis). Torsvik et al. state that their 1989 data were in in situ coordinates and the regional bedding correction for the shallow SW directed B component inferred by our alternate interpretation would make the magnetization dip approximately 50" downwards, which they contend is an unacceptable result. But Torsvik et al.'s (1989) preferred hightemperature direction for this sample with Dec. /Znc. of 222/28 transforms to c. 190/80 after correction for bedding tilt and is no less unacceptable. Torsvik et al. (1989) claim that their high-temperature remanence is pre-tectonic, but the direction arrived at for this sample is anomalously steep even for the intermediately inclined palaeomagnetic axis they adhere to. They further argue that we applied improper analytical procedures in our analyses. First we fail to see the ...