Yet there is, after two hundred years of debate, surprisingly little direct evidence about the effects of more, more productive, or better distributed land or assets (overall or of particular types) on fertility and migration - let alone about knock-on effects on resource depletion and pollution.(see note 1) This study, made possible by a grant from the European Commission, starts to provide such evidence. Specialists from several disciplines (economics, demography, sociology and statistics), representing research institutions in five countries, combined to design common questionnaires, exploring current and past assets, fertility, migration, and related issues at household level. After piloting and amendment, these questionnaires were then administered in 1999-2000 to samples of about 500 households in 25 villages in each of three roughly comparable agro-ecologies: rural drylands, with cattle and crops, in Rajasthan (India), Northern Province (South Africa), and Botswana.

Data cleaning was completed in 2000 and analysis has been proceeding since then. The following drafts of three country papers, and two of the four topic papers (on assets and fertility), represent provisional outputs and findings. We warmly welcome comments (preferably direct to the paper authors) but request that the papers should not be quoted, nor the findings used, at present.

We found that greater availability of land and other assets to poorer, less asset-endowed persons or households (whether through growth or redistribution) is associated with, and probably helps cause, somewhat reduced fertility at community and household level. The fertility effect is less than that of education or of realised female autonomy, but these too may be enhanced by greater access to land, assets and income. Contrary to expectations, we did not find that such access has an offsetting migration effect, slowing out-migration. Hence assets for the poor tend to reduce "population pressure". Other research suggests that such reductions (and associated rising worker /dependent ratios) bring faster, more equitable economic growth, but our work does not suggest that they greatly affect drylands environmental practice, experience, or resource depletion. Interesting incidental results emerged, e.g. that despite the usual account of rural households as "bricoleurs" with livelihoods from many sources, 90 per cent of the South African sample relied heavily (typically for over 80 per cent of income) on migrancy, pensions, or local incomes alone. Given the interest and likely debate around such findings, we think it best to share them, for comments, at once via these web pages. But please remember that these are provisional results. Careful readers will notice errors and omissions - and some inconsistencies, reflecting differences of interpretation or opinion. Some, but not all, will be clarified as work proceeds. In particular, however, some items require special caution:

• Not all households in the samples had the same probability of selection, in part because three villages in each country were deliberately sampled more densely. Also, there the calculation of adult equivalents is not yet complete and standardised in all papers. Further, the valuation of land and livestock is under review. The paper on assets (Eastwood and Lipton) corrects for all these items, but so far deals only with the S African data set. The paper on fertility (de Santis, Salvini et al.) and the S African country paper (Kirsten et al.) correct for some of these items. As versions of papers correcting for all items become available, they will be used to replace the versions on these web pages.
• The interpretations used to derive the Lorenz curves, Gini distributions, and other measures of asset inequality in the Botswana paper may require correction and are under review.