Science: So farewell, snowy owl: official - How the maps were compiled

The method chosen had to be standardised, straightforward and reliable. Fieldwork was organised using the national grid's 10km by 10km squares. Each of these - nearly 4,000 in Britain and Ireland - is divided on Ordnance Survey maps into 1km by 1km squares. Groups of four are called tetrads and there are 25 of them in each 10km by 10km square.

Observers, almost all volunteers, had to visit at least eight tetrads in each square, and spend exactly two hours in spring scouring habitats to compile a list of bird species. Without the time limit, maps derived from these counts would have reflected abundance of bird-watchers rather than birds, since southern England is bursting with amateur recorders while County Donegal or Caithness are not.

In this way an index of abundance was built up, based on the presence of each species in a known proportion of tetrads in squares country-wide. A pilot survey had shown that such an index is correlated with the real density of a species. It does not then matter that some bird species are more or less conspicuous than others, so long as the shy ones play as hard to find in Cumbria as they do in Kent. Slightly different methods had to be used for birds that nest in colonies, such as some sea birds, herons and sand martins. More than 1.5 million records were collected from nearly 43,000 tetrads.

Coloured abundance maps were drawn by computer for each breeding species, the first time such maps have been constructed in Britain. Areas of greatest abundance are coloured deep red, shading through red, orange, yellow and green to pale blue and white, as the species becomes more sparse - or absent.

Maps showing distribution of species (rather than abundance) are based on dots indicating the presence or absence of a species in each 10km square. These maps were compiled by adding to the two-hour tetrad data all records of breeding birds collected in each 10km square without a time limit. A third map for each species shows the change in "dot" distribution between the current (1988-1991) census and the previous one (1968-1972).

(Photograph omitted)