High precision for radiocarbon cannot be reached without profound insight into the various sources of uncertainty which only can be obtained from systematic investigations. In this paper, we present a whole series of investigations where in some cases 16O:17O:18O served as a substitute for 12C:13C:14C. This circumvents the disadvantages of event counting, providing more precise results in a much shorter time. As expected, not a single effect but a combination of many effects of similar importance were found to be limiting the precision.
We will discuss the influence of machine tuning and stability, isotope fractionation, beam current, space charge effects, sputter target geometry, and cratering. Refined measurement and data evaluation procedures allow one to overcome several of these limitations. Systematic measurements on FIRI-D wood show that a measurement precision of ±20 14C yr (1 σ) can be achieved for single-sputter targets.