A measured radiocarbon concentration equal to 80% of the conventional modern reference corresponds to roughly 1,793 radiocarbon years BP under the standard age equation. It does not follow that the sample must be from the calendar year 157 CE. The ‘BP’ clock is a laboratory convention anchored to 1950, while atmospheric radiocarbon, reservoir effects and calibration curves determine which real calendar dates the measurement can support.
This guide explains what radiocarbon instruments observe, how an uncalibrated result is calculated, why the traditional 5,568-year half-life remains embedded in reporting despite the more accurate physical half-life near 5,730 years, and how calibrated probability ranges replace the tempting but often false ‘1950 minus BP’ shortcut.
Find the question you need
Read from beginning to end or choose the dating question you need. All chapters stay open; the grouped index is expandable.
Understand the radiocarbon clock
Chapters 1–3. Start this route.
Calculate and calibrate
Chapters 4–6. Start this route.
Interpret uncertainty and reservoirs
Chapters 7–10. Start this route.
Practise and report
Chapters 11–12. Start this route.
Chapter index · Practice with answers · Geochronology subject root
Chapter index
Understand the radiocarbon clock · Chapters 1–3
Calculate and calibrate · Chapters 4–6
Interpret uncertainty and reservoirs · Chapters 7–10
Practise and report · Chapters 11–12
Radiocarbon dating concerns carbon that participated in an environmental carbon cycle before an organism died or its carbon exchange stopped. Radioactive ¹⁴C is present in carbon reservoirs, and its abundance subsequently changes through decay and sometimes through exchange, contamination or mixing.
A laboratory may measure isotope ratios using accelerator mass spectrometry or beta activity in other methods. The output is processed against standards, fractionation conventions and blanks before a conventional radiocarbon age is reported. The instrument does not directly observe a historical calendar year.
The measured carbon must relate to the archaeological event under investigation. Charcoal from a reused old timber may predate a fireplace, and a bone sample may be disturbed by later conservation chemicals. The age of the carbon is not automatically the age of every artifact or event found nearby.
This distinction mirrors other chronometers: a number is a measurement-based age under assumptions, whereas a claim about an event requires contextual evidence. The calibration step corrects a radiocarbon time scale; it does not cure every problem of sample selection or contamination.
Contents · Next · Subject root
Conventional radiocarbon reports use BP, meaning ‘before present’, with the reference ‘present’ fixed at AD 1950. It is not redefined every time a laboratory publishes results or every time the calendar advances. A radiocarbon age of 1000 BP in 2026 is not automatically ‘1000 years before 2026’.
One reason for the fixed reference is comparability. Published determinations retain one reporting baseline regardless of which year a measurement was made. This allows later researchers to calibrate or reinterpret results without silently changing their time coordinate.
Do not confuse uncalibrated ¹⁴C BP with calibrated calendar cal BP. Both use 1950 as a reference, but the first describes conventional radiocarbon age and the second denotes calendar-year intervals obtained after calibration.
The distinction also matters when dates near the BCE/CE boundary are discussed. An uncalibrated BP figure should not be mechanically subtracted from 1950 to give a precise historical year; that operation ignores the calibration step.
Contents · Previous · Next · Subject root
For a simplified conventional calculation, let F be the appropriately normalised fraction of modern radiocarbon relative to the reference. Write t₁₄C=−8033 ln F in radiocarbon years BP. The constant 8033 years comes from the conventional Libby half-life of about 5568 years divided by ln2.
This expression is a reporting convention, not a claim that atmospheric radiocarbon was constant through prehistory. The age behaves logarithmically: F=1 gives zero conventional age, and F=1/2 yields approximately 5568 BP.
Because F is dimensionless, the logarithm is valid. Using 80 rather than 0.80 in the equation would give a negative age of large magnitude and indicate a percentage-to-fraction error. Similarly, a calculator using base-10 LOG rather than natural LN without changing the factor gives the wrong result.
Real conventional-age practice includes fractionation and reference-standard definitions; our F examples are deliberately idealised. For laboratory interpretation use the complete measured result, its error and relevant correction metadata.
Contents · Previous · Next · Subject root
Take F=0.80. Natural logarithm ln(0.80)≈−0.22314355. Multiply by −8033 years to obtain t₁₄C≈1792.5 BP, usually rounded to about 1793 BP in the absence of stated measurement precision.
An arithmetic conversion 1950−1793≈AD157 is only what one would obtain by pretending an uncalibrated radiocarbon age were identical to elapsed calendar time. It is not a calibrated conclusion and should not be published as the specimen’s assigned calendar year.
To calibrate the measurement, compare its conventional radiocarbon age and uncertainty with a suitable radiocarbon calibration curve, taking geographic reservoir and hemisphere assumptions into account. The resulting calendar interpretation may cover a range of years and sometimes multiple separated intervals.
This distinction protects readers from false precision. One measured fraction plus the mathematical reporting equation is enough to compute a conventional age, but not enough to claim the actual depositional, death or construction event at one particular CE date.
| F (illustrative) | Conventional radiocarbon age | Calendar conclusion? |
|---|---|---|
| 1.00 | 0 BP | Not a full contextual date |
| 0.80 | ≈1793 BP | Requires calibration |
| 0.50 | ≈5568 BP | Requires calibration |
| 0.25 | ≈11136 BP | Requires calibration |
Contents · Previous · Next · Subject root
The physical half-life of radiocarbon is approximately 5730 years, while the historical Libby half-life of 5568 years remains part of the conventional reporting definition. It is therefore possible to see both numbers in reliable scientific documents without one being an accidental typographical error.
The conventional transformation constant 8033 years reflects 5568/ln2. In an ideal closed model using the more accurate physical half-life, a 50% fraction would instead correspond to about 5730 elapsed years. These two calculations answer different modelling or reporting questions.
One must not ‘correct’ a published 14C BP date by multiplying it blindly by 5730/5568 and then announcing a true calendar date. Historical atmospheric radiocarbon concentrations varied, so empirical calibration—not just replacement of one half-life—is required.
Explain the convention when teaching the topic. It is a powerful example of how a scientific reporting standard may retain a historical constant for comparability while a physical parameter has been measured more accurately.
Contents · Previous · Next · Subject root
Atmospheric radiocarbon abundance has changed over time. Organisms forming at different calendar dates did not always begin with identical ¹⁴C levels, so the uncalibrated radiocarbon-time scale can be warped relative to the historical calendar.
Calibration curves use independently dated archives, prominently tree rings for certain periods, together with other records and statistical methods for older intervals. The IntCal20 Northern Hemisphere atmospheric curve, published in 2020, is one well-documented standard data product, with a calibrated extent to 55,000 calendar years BP.
Calibration is not the act of adding or subtracting one universal constant. The relationship between radiocarbon BP and cal BP is not always one-to-one or even locally monotonic in a simple way. Wiggles and plateaus in the calibration curve can spread a radiocarbon determination over different calendar possibilities.
State the curve version used, not just ‘calibrated’. A change of calibration dataset can revise inferred date ranges while leaving the original laboratory measurement untouched, particularly in intervals where new reference data alter the curve.
Contents · Previous · Next · Subject root
Suppose a laboratory reports an idealised conventional result such as 1800±30 BP. The ±30 expresses uncertainty on the radiocarbon measurement under the laboratory’s error convention; it is not automatically a ±30-calendar-year range after calibration.
Calibration combines the measurement likelihood with the calibration curve’s time-varying relationship and uncertainty. The result may be reported as one or several calendar ranges with associated probability. A 95% calibrated range is therefore not necessarily centred on the naive calendar subtraction.
If a curve has a plateau, a small change in the radiocarbon result can correspond to a wide calendar interval. If several calendar ages share similar radiocarbon values, separate probability peaks can emerge. Reporting only a single best calendar date conceals that structure.
A clear article distinguishes an uncalibrated determination, the calibration curve and its version, and the reported calendar probability interval. The notation ‘cal BP’ or calibrated BCE/CE dates helps prevent accidental mixing with ordinary BP values.
Contents · Previous · Next · Subject root
Some organisms draw carbon from reservoirs that differ in radiocarbon activity from the atmosphere. Marine organisms can incorporate carbon whose apparent radiocarbon age differs from an equivalent-age terrestrial organism. Freshwater systems can also exhibit hardwater or old-carbon effects.
For this reason, a marine shell and a terrestrial seed from the same archaeological context can yield different conventional radiocarbon determinations even if both materials were formed around the same time. The discrepancy is not automatically analytical incompetence or evidence of two separate human occupations.
Marine calibration may use an appropriate marine curve and locally researched reservoir offset with uncertainty. Freshwater cases can demand site-specific investigation, since carbon sources and pathways can vary substantially.
Do not invent a universal ‘400-year marine correction’ for every coast and time. Reservoir ages vary across locations and periods, and a correction suitable for one setting may be misleading in another. The selected material and its ecology belong in the chronology explanation.
Contents · Previous · Next · Subject root
Adding a small quantity of modern carbon to an old sample can shift its measured radiocarbon result younger; introducing very old or fossil carbon can make it appear older. Effects are especially serious when the genuine surviving ¹⁴C fraction is small and background or contamination becomes a large part of the signal.
Laboratories pretreat samples to remove relevant contaminants, use standards and blanks, and assess whether sufficient datable material survives. The specific procedure depends on the material, such as charcoal, bone collagen, shell carbonate or textile fibres.
Calibration cannot know that a conservation consolidant was added to a museum object unless the contamination is identified and handled at the sample-preparation or interpretation stage. An apparently narrow calibrated date can still be wrong for the archaeological event if the wrong carbon was measured.
Good provenance includes sample type, pretreatment, context, contamination screening and any independent age constraints. This is why ‘calibrated’ should not be used as a synonym for ‘certain’.
Contents · Previous · Next · Subject root
Radiocarbon activity decreases exponentially, so after many half-lives very little ¹⁴C remains. Oxford’s accelerator unit describes an approximate practical method limit around 50,000–55,000 years, with interpretation constrained by background and sample condition.
This is not a cliff at which every sample younger than 50,000 years is perfectly datable and every older one is impossible to assess. Actual detection capability, contamination, uncertainty and calibration coverage must be considered.
At a sufficiently old age, small modern-carbon contamination can become comparable to the surviving authentic signal. A determination near an instrument’s background limit may be best interpreted as a minimum age or a result requiring special treatment, rather than an exact calendar year.
Other chronometers such as uranium-series or luminescence methods can constrain contexts outside radiocarbon’s effective range, but they date different physical processes and have their own assumptions. Choice of method should follow the question and the material.
Contents · Previous · Next · Subject root
Exercise A: Under the conventional equation, F=0.50 gives t≈5568 BP. The physically more accurate half-life around 5730 years is not a justification for replacing the conventional number inside a reported ¹⁴C BP label.
Exercise B: F=0.25 gives t≈11136 BP because the sample has gone through two conventional half-lives. Does that directly establish a calendar date 11136 years before 1950? No; use a suitable calibration curve and uncertainty model.
Exercise C: An uncalibrated sample reports 1800±30 BP. Can the parent write ‘AD150±30’ on a historical timeline? No. The calendar distribution requires calibration, potentially producing wider or multiple intervals.
Exercise D: A marine shell seems centuries older than associated terrestrial charcoal. Name at least two possible checks: reservoir effects and their local uncertainty, and whether either sample’s carbon actually records the same event.
Contents · Previous · Next · Subject root
Begin by describing the sample and event it might date. Verify its origin, material type, archaeological context and pretreatment, then report the laboratory conventional radiocarbon determination with its uncertainty and identifier where appropriate.
Next choose a defensible calibration curve or reservoir model, noting hemisphere, marine or terrestrial character, calibration version and any local offsets. Report calibrated date ranges with their probability rather than one unsupported subtraction from 1950.
Finally compare the result with independent archaeological or environmental evidence, and state where sample selection or contamination could shift interpretation. A scientifically useful chronology exposes its assumptions so later calibration revisions can be evaluated.
Continue with The Gold Standard of Geochronology and the marine reservoir correction guide. These lessons distinguish measured isotope age, calendar calibration and the geological event being dated.
Applied radiocarbon chronology lab
Uncertainty propagation for the idealised F example
From t=−8033 lnF, the derivative is dt/dF=−8033/F. If the illustrative F is 0.800 with standard uncertainty 0.008, the first-order radiocarbon-age uncertainty is about 8033×0.008/0.800=80 years.
The nominal conventional result is approximately 1793±80 BP under the simplified arithmetic assumptions. This is not yet a calibrated calendar probability range; it represents only propagated uncertainty in the conventional age from the chosen hypothetical F.
Laboratory uncertainties can contain additional components such as background and normalisation. A genuinely calibrated distribution also incorporates the calibration curve and any relevant reservoir-model uncertainty.
How a calibration plateau can widen a calendar interpretation
Imagine, purely as a teaching sketch, that a calibration curve stays near the same radiocarbon BP value across a broad calendar period. One measured age may then be compatible with many real calendar dates, even when its laboratory uncertainty is modest.
The outcome is not an error caused by rounding. It is a consequence of the inverse problem being weakly identified by the available radiocarbon measurement in that part of the curve.
Do not invent exact BCE/CE intervals without running a documented calibration on the supplied ¹⁴C age. A picture of a plateau demonstrates the geometry of the inference, but the reported calendar ranges require actual curve data.
A matched-pair interpretation exercise
An archaeological deposit contains a charred seed and a marine shell. The shell returns an older conventional radiocarbon age. Is the shell necessarily older than the seed? Not without checking reservoir effects, the ecology of the organism, pretreatment and contextual association.
A defensible comparison records both measured ages, their uncertainties and appropriate calibration/reservoir models. A raw subtraction of the shell’s age from the seed’s age can misstate the chronological difference.
This illustrates the general rule that two dates obtained from unlike carbon reservoirs should not be compared on an uncalibrated basis as though they share the same initial isotope activity.
Teach three separate clocks on one timeline
Draw a line labelled calendar years and mark 1950 as the cal BP zero. Below it show an uncalibrated radiocarbon-age axis, explicitly identified as a reporting time scale rather than a perfect translation of the calendar.
Ask learners to place a hypothetical 1800 BP radiocarbon determination on the lower axis and explain why its placement does not uniquely define a point on the calendar line above. They should identify what calibration data and reservoir knowledge would be needed.
The exercise helps students understand that reference-year conventions, physical decay and empirical calibration are related but separate parts of chronological inference.
Frequently asked questions
Does BP mean years before 2026?
No. In radiocarbon convention the ‘present’ reference is fixed at 1950. The reporting reference does not change as calendar years pass.
Can I subtract BP from 1950 to get CE?
Not for an uncalibrated ¹⁴C BP determination. Atmospheric changes and other effects require calibration to derive calendar ranges.
Is 5568 or 5730 the correct half-life?
Both appear for different reasons: 5568 years is built into the historical conventional reporting calculation, while approximately 5730 years is the more accurate physical half-life.
Sources and related reading
The numerical fractions and uncertainty example are constructed for teaching; no specific archaeological specimen has been dated here.
- Oxford Radiocarbon Accelerator Unit — Radiocarbon Dating and Reporting
- Oxford Radiocarbon Accelerator Unit — Radiocarbon Calibration
- Cambridge University Press — IntCal20 Northern Hemisphere Calibration Curve
- Palaeoclimatology | Marine Reservoir Corrections and Calendar Years
Advanced chronological lab: uncertainty, reservoirs and reporting
Recalculate conventional age without mixing years and isotope fractions
For a conventional fraction of modern F=0.80, the calculation is t=−8033 ln(0.80)≈1792.5 radiocarbon years BP. If another person writes F=80 in a spreadsheet, the natural logarithm changes sign and the output becomes meaningless for the intended fraction. Record F as 0.80, not eighty, and label it dimensionless.
An easy reverse check begins with t=5568 conventional BP and solves F=exp(−t/8033). The result is approximately one-half. For t=11136 conventional BP, the same inverse returns one-quarter. The equations are consistent when time is expressed in years and the exponential and logarithm use matching bases.
These reversals verify the conventional mathematical calculation, not the actual sample’s calendar age. The difference between a good numerical solution and a good chronological interpretation begins exactly here: laboratory reporting conventions do not reproduce every historical atmospheric condition.
Students should keep three fields separate: measured fraction with its uncertainty, conventional radiocarbon age in BP, and calibrated calendar interpretation. When a result is copied into a timeline, its field label and reference curve must travel with it.
Propagate fraction uncertainty before invoking a calibration curve
Differentiate t(F)=−8033 lnF to obtain dt/dF=−8033/F. For the constructed value F=0.800±0.008, the first-order magnitude of the conventional-age uncertainty is about (8033/0.800)×0.008≈80 years. A simplified statement is therefore about 1793±80 radiocarbon BP.
The sensitivity is proportional to 1/F. As the remaining radiocarbon fraction becomes smaller, the same absolute error in F produces a larger age uncertainty. This is one reason that older samples and low-carbon backgrounds require particularly careful laboratory treatment.
That propagated ±80 is not a calendar-year confidence interval. A calibrated distribution also depends on the curve, its uncertainty, reservoir considerations and the shape of the relationship between ¹⁴C age and calendar time.
A sophisticated report should not apply a mechanical ‘±’ conversion after calibration if the resulting distribution is asymmetric or multimodal. Quoted calendar ranges and probability masses should come from the documented calibration method.
Distinguish reservoir offset from contamination with old carbon
Marine reservoir differences arise because organisms can assimilate carbon with a radiocarbon activity history different from the atmosphere. Freshwater hardwater effects can result when an aquatic food web includes dissolved carbon from old carbonate sources. These are physical characteristics of how the carbon was acquired.
Contamination describes the addition or mixing of carbon that is not part of the intended sample’s original signal, perhaps during burial, conservation, handling or laboratory preparation. Both may shift an apparent radiocarbon result, but the mechanisms and suitable corrections are different.
A shell’s apparent age relative to associated terrestrial charcoal should therefore motivate investigation of ecological carbon source, local reservoir data, pretreatment, material preservation and whether the two samples actually record one event. Subtracting a single universal correction from every marine shell is not scientifically justified.
Chronological interpretation also requires contextual judgement. A charred timber can be older than the fire that eventually burned it, even if the radiocarbon measurement and calibration are otherwise sound. The age of sampled carbon is not automatically the age of its last use.
A reporting template that survives calibration revisions
A reusable chronology entry contains the lab identifier; sample material and archaeological context; pretreatment; conventional radiocarbon age with its laboratory uncertainty; calibration curve and version; reservoir model if relevant; and the resulting calendar probability range or ranges.
When calibration methods or reference curves are updated, retain the original uncalibrated determination. Future researchers can then repeat the calibration using improved data without reconstructing a lost laboratory measurement from a previously rounded calendar date.
The published IntCal20 Northern Hemisphere curve is explicitly a calibrated reference for appropriate terrestrial samples. Use of a Southern Hemisphere or marine dataset requires matching the underlying carbon reservoir and the research question, not merely choosing whichever curve yields a preferred historical date.
Good educational prose makes the uncertainty intelligible to non-specialists: ‘The carbon measurement supports these calendar ranges under a specified model’ is more accurate than ‘The object was made exactly in this year.’
A final distinction between sample age and event age
Imagine a wooden beam cut from a much older tree and reused in a building centuries later. Its radiocarbon signal reflects carbon fixed during tree growth, not necessarily the year the building was constructed. A perfectly calibrated date from interior wood can still predate the building event. Archaeological interpretation therefore needs timber anatomy, ring position and contextual evidence.
Calibration answers how a radiocarbon measurement relates to calendar time under a selected reference model. It does not establish that the sampled carbon was formed at the same time as the human action being investigated. Report both the calibrated ranges and why the sample is believed to date the event.