Radiocarbon dating with accelerators requires the manufacture of suitable carbon targets. Carbon yield and 13C fractionation were investigated for the simple and direct pyrolysis of wood and cellulose to charcoal. Under continuous vacuum removal of evolved volatiles, carbon yields of 35 to 40% and Delta-13C fractionation of −2.5 per mil were observed in the pyrolysis of wood to charcoal, whereas yields of 30% and fractionation of −0.8 per mil were obtained in the pyrolysis of cellulose to charcoal. Yield and fractionation leveled off at temperatures above 300 degrees C. Yields and fractionations were also measured for pyrolysis of wood and cellulose in a continously-flowing argon atmosphere. Yields were higher and fractionations smaller than for the corresponding vacuum cases. For cellulose sealed in evacuated glass tubes and pyrolized at 550 to 600 degrees C, carbon yields greater than 60% and fractionation of about −0.5 per mil were observed. Yields increased and fractionation tended to decrease as the ratio of tube volume/mass of cellulose decreased, ie, as the pressure increased. Reheating of this charcoal under continuous vacuum pumping revealed no loss of mass and no alteration of carbon isotopic composition. Fractionation measurements were additionally performed on wood and the charcoal produced from burning in a fireplace, conditions approximating the “natural“ production of charcoal. Despite the large potential fractionation suggested in the wood pyrolysis experiments, charcoal produced in the fireplace showed very small or no fractionation.

A high-precision calibration curve, derived from the radiocarbon age determinations of 195 decade samples spanning the AD 1 to 1950 interval, is presented. Though derived for the Pacific Northwest and California, the curve can be used for a large part of the northern hemisphere. This is proven by the radiocarbon ages of contemporaneous sample pairs which are, in most instances, identical within the quoted precision. Two sets of single-year data reveal no evidence for an 11-year cycle with an amplitude beyond the 12-year measuring precision. This indicates that the calibration curve is also applicable for single-year 14C samples. Analysis of the Seattle data sets and comparison with those published by the Belfast, La Jolla, and Heidelberg laboratories show that the total variability in a radiocarbon age determination is often larger than that predicted from the quoted errors. Upper limits for the error multiplier (ie, the factor with which the quoted error has to be multiplied to obtain the overall laboratory variability) are estimated at 1.5 for Seattle and Belfast, 1.1 to 1.4 for La Jolla, and 2.0 for Heidelberg. The comparisons with Belfast, La Jolla, and Heidelberg also reveal offsets with the Seattle calibration curve of, respectively, 4, 27 to 55, and 58 years. These offsets are most likely due to laboratory bias. An improvement of the present calibration curve by combining data sets from other laboratories will only be possible when offsets and error multipliers are precisely known through interlaboratory calibration.