The last of these reports appeared four years ago; 1 a number of major advances in structure elucidation have appeared in laboratories in that period, mostly in the ¢eld of macromolecular structure determination. This is largely because sub-100 non-hydrogen atom structure solution has become almost routine, with extremely robust solution and re¢nement programs being complemented by rapid data collection from single crystal samples. In many cases, the role of the in-house crystallographer in chemistry departments is to provide a service to synthetic chemists in an analogous way to that of NMR spectroscopists ¢fteen or twenty years ago. Many chemical crystallographers have used this as an opportunity to study more dif¢cult structures, for example larger molecules from both single crystal and powder methods, using twin datasets, and looking more closely at structures in charge density studies.This increased ease has not removed the perils of being ``Marshed''; the well-known investigator into errors in space group determination has examined structures deposited into the Cambridge Structural Database (CSD) with space group P1, of which more than 20% are incorrect. 2 The majority of errors appear to be due to typographical errors in the original publication, but a signi¢cant minority are due to incorrect space group assignment (usually authors missing a centre of inversion, and assigning P1 rather than P1, or not recognizing higher symmetry, e.g., assigning P1 (triclinic) rather than C2 (monoclinic). However, the author notes that in some cases of true P1 structures, the deviation from centrosymmetry is as small as 0.1 Ð, and would be expected to lead to dif¢culties in re¢nement unless the problem had been recognized, as the following example illustrates.The solution and re¢nement of the antibiotics Emycin E and D provides a warning not to abide by conventional wisdom under all circumstances. 3 Both antibiotics crystallize in a triclinic space group with two molecules in the asymmetric unit (Z 2). The dataset for Emycin E was in accordance with the normal measure of centricity, a mean hjE 2 À 1ji value of 0.978 (the expected value for a centrosymmetric dataset is 0.968, and is 0.736 for acentric), while that for Emycin D was 0.829. Both can be re¢ned to low residual indices (R-factors) in space group