Various timestamping mechanisms has been provided in order to characterize the partial order between events occuring in a distributed computation. Using the order induced by such timestamping mechanisms, one can reasonn about causal or concurrent relation-ships between events. However comparisons between timestamps can induce false assumptions on these relationships. The order induced by the timestamps is not always the existing partial one in a distributed execution. It is in these terms that we study the accuracy of three timestamping mechanisms, showing the type of compromise (space-accuracy tradeoffs) one has to make when using them.