The extensive use of the C
37 di- andtriunsaturatedalkenones for the estimation of sea surface temperaturesin ancient oceans prompts an investigation of the analytical constraints on reliable gas chromatographic measurements. Several tests reproducing the alkenone amountscurrently encountered in high-resolution
stratigraphystudies show that no significant errors are specificallyrelated to differences between on-column, splitless, andseptum-programmable injection. However, irreversibleadsorption on the chromatographic column is one majorsource of error that has significant effects (temperaturedeviations higher than 0.5
C) when the total amounts ofalkenones introduced into the system are lower than5-10 ng. These deviations are more important forlowerrelative proportions of the triunsaturated species (highertemperature). The observed adsorption evidences that,in the practical operating conditions for analysis oflargedata sets (average of three replicate injections for 1 gofsample), a concentration of 50 ng/g is the lowest limitfor reliable paleotemperature estimation.