Radiocarbon dating involves a comparison of the specific

β

-activity of the sample (A) with that of the modern reference standard (A

0

) . The age of the sample is given by: T=82671n (A

0

/A) (year) . As the sample and the standard are prepared and counted in the same way

it is usually only the statistics of counting is taken into consideration. However

in practice

errors introduced in experimental techniques are not the same for sample and standard. Some foreseeable sources of errors in liquid scintillation counting are investigated in this paper. An overall precision in the activity-ratio A

0

/A of about 0.5% is achieved.