Radiocarbon dating is the closest thing archaeology has to a stopwatch, yet the number on the lab report almost never equals the object’s real age. A timber beam that carbon-dates to 2,400 years before present was not necessarily cut 2,400 years ago, and a bone fragment measuring 12,000 radiocarbon years may actually be closer to 14,000 calendar years old. The gap between the laboratory figure and the calendar date is not an error to be corrected — it is a known, measured, and published offset that every archaeologist has to apply before the number means anything.
What the lab actually measures
Living organisms absorb carbon from their environment in two forms: stable carbon-12 and a small fraction of radioactive carbon-14. While an organism is alive, the ratio between them stays roughly constant because it keeps exchanging carbon through food, air, or water. The moment it dies — a tree is felled, an animal stops breathing, a shell stops growing — uptake stops and the carbon-14 already inside begins to decay. Its half-life is 5,730 years, meaning half of the remaining carbon-14 is gone after that span, another half of what is left after 11,460 years, and so on.
The laboratory does not read a date. It measures the ratio of carbon-14 to carbon-12 in a sample, compares it to a modern reference standard, and computes how many half-lives must have elapsed to produce that ratio. That raw result is expressed as “radiocarbon years before present,” where “present” is conventionally fixed at 1950 AD — not today. A result written as 3,000 BP refers to 3,000 radiocarbon years before 1950.
Why radiocarbon years drift from calendar years
The stumbling block is that atmospheric carbon-14 is not constant over time. It varies with solar activity, with geomagnetic field strength, with volcanic eruptions, and with the burning of fossil fuels. A sample that died during a period when the atmosphere held more carbon-14 will look younger than it is, because it absorbed a richer starting mixture; a sample from a carbon-poor period will look older.
To solve this, laboratories use calibration curves built from independently dated material — tree rings counted year by year across thousands of years, corals dated by uranium-thorium methods, and lake sediments with annual layers. Each calibration curve maps a radiocarbon measurement onto a probability distribution of calendar years. Modern curves such as IntCal20 and SHCal20 (for the Southern Hemisphere) are updated as new reference sequences are published, which is why the same sample can be calibrated today to a result slightly different from the one printed on a 1990s report.
| Radiocarbon result | Approximate calibrated range | Why the gap grows |
|---|---|---|
| 500 BP | 530–490 cal BP (narrow) | Recent material, curve is well resolved |
| 3,000 BP | 3,200–3,050 cal BP | Plateau in the curve during the period |
| 12,000 BP | 14,000–13,800 cal BP | Large offset during deglaciation |
| 40,000 BP | 44,000–43,000 cal BP | Near the method’s practical limit |
Contamination: the error that calibration cannot fix
Calibration corrects a known systematic offset. Contamination is different — it introduces new carbon after death and quietly distorts the measurement itself. A sample that absorbed just 1% of modern carbon will read as dramatically younger than it is, and the effect grows worse with age. For a 10,000-year-old bone, 1% modern contamination pulls the result back by roughly 1,000 radiocarbon years; for a 40,000-year-old sample, the same 1% can halve the apparent age.
This is why laboratories pretreat samples so aggressively. Bone goes through collagen extraction, charcoal is washed in acid-base-acid sequences, and shell is checked for recrystallization. A sample submitted with visible rootlets, museum consolidant, or soil carbonate will often come back with a warning note, and a cautious archaeologist should treat that result as a maximum age rather than a date.
Where the method breaks down
Radiocarbon dating depends on the sample having been in equilibrium with the atmosphere. Marine organisms incorporate “reservoir” carbon that has been isolated from the atmosphere for centuries, so a fresh seashell can appear hundreds of years old. Freshwater fish that feed in limestone-rich watersheds face the same problem. Organisms eating marine protein carry a mixed signal that requires a dietary correction factor.
The method also runs out of precision beyond about 50,000 years, where remaining carbon-14 is too scarce to count reliably, and it loses resolution in the “radiocarbon plateau” periods where the calibration curve is nearly flat — several calendar centuries collapse into a single radiocarbon value, so a sample from such a period can only be placed as “somewhere in this 300-year window.”
Reading a dating report without misreading it
Three numbers appear on a properly written report and each means something different: the raw radiocarbon age (BP), the laboratory error term (±), and the calibrated calendar range with its confidence level. The error term is a statistical bound on the measurement, not on the true age — a 4,000 ± 30 BP result with a 300-year calibrated range is normal, because curve uncertainty dominates.
For anything that has to be placed on a real timeline, the calibrated range is the only number worth quoting. If you are working out how far back an event sits from a known calendar anchor — a structure’s construction phase, a family’s generational span, a historical era’s boundary — our historical era calculator converts a fixed anchor point into elapsed years, months and days, which is the same arithmetic a lab performs when it calibrates a sample against a reference sequence. When the anchor is a birth date or a document rather than a sample, the chronological age calculator gives the exact calendar interval without any radiocarbon uncertainty at all.
The short version
A radiocarbon date is a measurement of isotope ratios, expressed against a 1950 reference year, and it describes the atmosphere the organism once lived in rather than the calendar date of its death. Turning it into a usable date requires calibration to a published curve, an assessment of contamination and dietary reservoir effects, and an honest statement of the confidence range. Museums and journals increasingly reject “dated to 3,000 BC” written from an uncalibrated figure — because the number that came off the machine was never a calendar date in the first place.



