Erdil and colleagues compared, by intrathecal injection, plain solutions of racemic bupivacaine and levobupivacaine, both with fentanyl added to the injectate [1]. They found that the racemic bupivacaine solution spread more rapidly and produced a more extensive spinal anaesthetic block with greater incidences of both hypotension and nausea, findings that do not match those of any previous worker. They discuss a number of very minor issues (age of their patients, relative potency of the drugs, and their differential vasoconstrictor action) in an attempt to explain these results, but ignore completely the two most important factors affecting intrathecal drug spread: the density of the solution and the position of the patient [2].Lumbar puncture and drug injection were performed in the sitting position and I will take much convincing to change my view that the bupivacaine solutions used were less dense than those of levobupivacaine so that they 'floated' to a greater cephalad level. This, not a difference in any inherent drug property of two very similar agents, is a much more likely explanation of their results. Unfortunately, no figures for solution density are provided in the paper, nor even are the sources of the drugs specified, so it is impossible to assess the position or replicate the study. It is over a century since Barker first demonstrated the importance of solution density in spinal anaesthesia [3], and those researching this technique must recognise and acknowledge its importance.