Structure predictions allowing more than one molecule in the asymmetric unit. Erratum

In the paper by van Eijck & Kroon (2000) the fifth and sixth sentences in the third paragraph of §2.1 are incorrect. The correct paragraph is here reprinted in full. Table 6 also contains errors and thus is also reprinted below.

In our procedure the space-group symmetry and the number of independent molecules, G, are chosen in advance. A Cartesian axes system (x, y, z) is used, where the crystallographic a axis coincides with the x axis and the b axis is positioned in the x,y plane. To create a trial structure the components a_x, b_y and c_z are given random values between 4 and 25 Å. For an orthorhombic space group this is sufficient; higher symmetry has not been implemented. In monoclinic space groups the variation of the cell angle is allowed by assigning a random value (between 0 and a_x) to c_x. In triclinic space groups b_x is similarly chosen between 0 and a_x, and c_y takes a random value between 0 and b_y. So for these three crystal systems the number of lattice parameters to be determined is 3, 4 and 6, respectively.