ecological and the peace movements), or the actually existing history and possible future of the labour movement.

Finally, dominant ideologies need to be rescued from their conversion into theses, whether by proponents or opponents. They should be developed as hypotheses of empirical research. As far as I can tell, AHT are quite right in rejecting the idea that all-pervasive normative doctrines govern the behaviour of members of developed societies. But, again, it would be obscurantist to refrain from looking into the dominant ideologies. Here a comparative approach seems to be the most fruitful. In complex societies, what is can be most easily discovered through comparison with what exists or has existed elsewhere. In my own research I have been looking at how political ideologies have changed in Swedish electoral campaigns from 1928 to 1982. In functioning democracies, what is said and what is not said, what is appealing and what is regarded as a campaign blunder, tap important aspects of ideological power relations in complex societies. Since they have a behavioural component, election campaigns also seem more reliable than international opinion polls. Another promising route -doubtless not the only one -- is to look at the prevalence or absence and the historical trajectory of certain concepts or labels of identification. For instance, in Swedish parlance there has been no 'middle class' or 'middle estate' [Mittelstand] since about 1950: but there are 'bourgeois parties' and a 'workers' movement' (without a working class).

With all the respect due to The Dominant Ideology Thesis for its intelligence, erudition and sound scepticism of the past, my fundamental objection is that it is not silence which is now on the agenda, that serious analysis of ideology has to begin and is beginning. Let me end by expressing the hope that Abercrombie, Hill and Turner will bring their undeniable skills to bear on this task.

-178-