The Gold Standard of Archaeology | Reference-Material Lot Changes, Bridging Studies and Long-Term Laboratory Comparability

1. Define the reference material’s job

Back to contents

A reference material may verify accuracy, monitor day-to-day control, support calibration, test extraction, train analysts or help compare laboratories. These roles require different evidence. A material suitable for quality control is not automatically a calibrator, and a pure solution may not challenge an archaeological matrix.

Before changing lots, state the measurand, method, matrix, concentration range, decision threshold and claim the material supports. Read the certificate and instructions rather than assuming that identical product names imply identical intended use.

First-principles check: State the decision this chapter protects, the evidence needed for it and the limit that remains. Write the rule before looking at the preferred outcome. Then identify the authority, observation or calculation that could change the answer. This keeps the method tied to evidence rather than hindsight.

Worked practice: Write intended-use statements for a control sample, calibrator and matrix reference material. Keep the intermediate steps visible. A reader should be able to locate the input, apply the same rule and understand why a different result would follow if one fact changed.

Changed-context transfer: The new lot is certified for concentration but not for the isotope ratio used locally. Preserve the underlying principle but replace the setting-specific details. State which parts travel, which do not, and which new authority or measurement would be needed.

Evidence record: Keep intended use, measurand, matrix, range, certificate version, restrictions and owner. Retain uncertainty and disagreement instead of compressing them into a clean label. The record should let another person reconstruct the decision without relying on the memory or status of the original decision-maker.

Decision-quality drill: Run the practice once under the most favourable plausible assumptions and again under the least favourable plausible assumptions. Compare which conclusion changes, which merely becomes less precise and which survives. If the result depends on an unstated convention, make that convention visible before acting. This is a transfer test: the method should guide a new case, not only explain the example after the answer is known.

Practical verification: Give the case file to a reviewer who did not build the analysis. Ask that person to reproduce the critical step, identify the stopping rule and name the strongest live alternative explanation. Resolve differences by returning to the governing evidence, not by averaging opinions. Record corrections, version the decision and keep a short list of conditions that would require reopening it.

2. Compare certificates before comparing measurements

Back to contents

Review assigned values, uncertainties, traceability, homogeneity, stability, minimum sample mass, storage, preparation, expiration and any changed analytical basis. A small difference in central value can be important near a decision limit; a larger difference may be harmless if results are normalised correctly and uncertainty is propagated.

Do not treat certificate uncertainty as the laboratory’s repeatability. It is one component in a larger measurement model. Record whether a new lot changes nominal value, uncertainty or use instructions before a single bridging result is examined.

First-principles check: State the decision this chapter protects, the evidence needed for it and the limit that remains. Write the rule before looking at the preferred outcome. Then identify the authority, observation or calculation that could change the answer. This keeps the method tied to evidence rather than hindsight.

Worked practice: Build a side-by-side certificate table and flag every changed field. Keep the intermediate steps visible. A reader should be able to locate the input, apply the same rule and understand why a different result would follow if one fact changed.

Changed-context transfer: The assigned value is stable but minimum sample mass doubles. Preserve the underlying principle but replace the setting-specific details. State which parts travel, which do not, and which new authority or measurement would be needed.

Evidence record: Retain certificates, comparison table, interpretation, open questions and approval. Retain uncertainty and disagreement instead of compressing them into a clean label. The record should let another person reconstruct the decision without relying on the memory or status of the original decision-maker.

Decision-quality drill: Run the practice once under the most favourable plausible assumptions and again under the least favourable plausible assumptions. Compare which conclusion changes, which merely becomes less precise and which survives. If the result depends on an unstated convention, make that convention visible before acting. This is a transfer test: the method should guide a new case, not only explain the example after the answer is known.

Practical verification: Give the case file to a reviewer who did not build the analysis. Ask that person to reproduce the critical step, identify the stopping rule and name the strongest live alternative explanation. Resolve differences by returning to the governing evidence, not by averaging opinions. Record corrections, version the decision and keep a short list of conditions that would require reopening it.

3. Plan overlap before the old lot is exhausted

Back to contents

The strongest bridge measures old and new lots during an overlap window under the same method, instruments, analysts and environmental conditions. Reserve enough old material for replication, troubleshooting and possible reruns. Randomise run order and distribute comparisons across days so a lot effect is not confounded with one batch.

If no overlap remains, the laboratory may use retained quality-control data or a stable third material, but the inference is weaker. Name that limitation instead of manufacturing a direct comparison from noncontemporaneous means.

First-principles check: State the decision this chapter protects, the evidence needed for it and the limit that remains. Write the rule before looking at the preferred outcome. Then identify the authority, observation or calculation that could change the answer. This keeps the method tied to evidence rather than hindsight.

Worked practice: Calculate the minimum old-lot reserve for three days, two instruments and duplicate preparation. Keep the intermediate steps visible. A reader should be able to locate the input, apply the same rule and understand why a different result would follow if one fact changed.

Changed-context transfer: Only archived control charts survive; no physical old material remains. Preserve the underlying principle but replace the setting-specific details. State which parts travel, which do not, and which new authority or measurement would be needed.

Evidence record: Keep bridge plan, reserve calculation, randomisation, run schedule, deviations and fallback evidence. Retain uncertainty and disagreement instead of compressing them into a clean label. The record should let another person reconstruct the decision without relying on the memory or status of the original decision-maker.

Decision-quality drill: Run the practice once under the most favourable plausible assumptions and again under the least favourable plausible assumptions. Compare which conclusion changes, which merely becomes less precise and which survives. If the result depends on an unstated convention, make that convention visible before acting. This is a transfer test: the method should guide a new case, not only explain the example after the answer is known.

Practical verification: Give the case file to a reviewer who did not build the analysis. Ask that person to reproduce the critical step, identify the stopping rule and name the strongest live alternative explanation. Resolve differences by returning to the governing evidence, not by averaging opinions. Record corrections, version the decision and keep a short list of conditions that would require reopening it.

4. Select matrices and levels that reveal non-equivalence

Back to contents

One concentration in a clean solution cannot establish comparability across bone, charcoal, sediment, metal or ceramic extracts. Choose materials that represent the method’s major matrix challenges and the range where decisions are made. Include blanks and low-level samples when contamination or background correction matters.

The goal is not to prove identity in every possible condition. It is to show whether the change is acceptable for defined uses. Prespecify allowable difference using scientific and decision relevance, not merely whether a statistical test fails to reject equality.

First-principles check: State the decision this chapter protects, the evidence needed for it and the limit that remains. Write the rule before looking at the preferred outcome. Then identify the authority, observation or calculation that could change the answer. This keeps the method tied to evidence rather than hindsight.

Worked practice: Choose a bridge panel for a method spanning low, medium and high archaeological concentrations. Keep the intermediate steps visible. A reader should be able to locate the input, apply the same rule and understand why a different result would follow if one fact changed.

Changed-context transfer: Add a matrix with unusually high organic interference. Preserve the underlying principle but replace the setting-specific details. State which parts travel, which do not, and which new authority or measurement would be needed.

Evidence record: Retain panel rationale, levels, matrix coverage, acceptance limits and unsupported uses. Retain uncertainty and disagreement instead of compressing them into a clean label. The record should let another person reconstruct the decision without relying on the memory or status of the original decision-maker.

Decision-quality drill: Run the practice once under the most favourable plausible assumptions and again under the least favourable plausible assumptions. Compare which conclusion changes, which merely becomes less precise and which survives. If the result depends on an unstated convention, make that convention visible before acting. This is a transfer test: the method should guide a new case, not only explain the example after the answer is known.

Practical verification: Give the case file to a reviewer who did not build the analysis. Ask that person to reproduce the critical step, identify the stopping rule and name the strongest live alternative explanation. Resolve differences by returning to the governing evidence, not by averaging opinions. Record corrections, version the decision and keep a short list of conditions that would require reopening it.

5. Separate lot, run, instrument and preparation effects

Back to contents

A bridge can confound the lot with the day, analyst, instrument drift or sample preparation. Use paired or blocked measurements where possible and include routine controls. Replicate independent preparation, not only repeated reading of one solution, when preparation is part of the measurement process.

Model the hierarchy that matters. An apparent two-percent lot shift that vanishes after instrument recalibration is not a lot correction. Conversely, averaging across instruments can hide a lot-by-platform interaction that affects only part of the laboratory.

First-principles check: State the decision this chapter protects, the evidence needed for it and the limit that remains. Write the rule before looking at the preferred outcome. Then identify the authority, observation or calculation that could change the answer. This keeps the method tied to evidence rather than hindsight.

Worked practice: Design a balanced bridge that estimates lot, day and instrument contributions. Keep the intermediate steps visible. A reader should be able to locate the input, apply the same rule and understand why a different result would follow if one fact changed.

Changed-context transfer: One instrument receives maintenance halfway through the comparison. Preserve the underlying principle but replace the setting-specific details. State which parts travel, which do not, and which new authority or measurement would be needed.

Evidence record: Keep run order, analysts, instruments, preparations, controls, maintenance and model specification. Retain uncertainty and disagreement instead of compressing them into a clean label. The record should let another person reconstruct the decision without relying on the memory or status of the original decision-maker.

Decision-quality drill: Run the practice once under the most favourable plausible assumptions and again under the least favourable plausible assumptions. Compare which conclusion changes, which merely becomes less precise and which survives. If the result depends on an unstated convention, make that convention visible before acting. This is a transfer test: the method should guide a new case, not only explain the example after the answer is known.

Practical verification: Give the case file to a reviewer who did not build the analysis. Ask that person to reproduce the critical step, identify the stopping rule and name the strongest live alternative explanation. Resolve differences by returning to the governing evidence, not by averaging opinions. Record corrections, version the decision and keep a short list of conditions that would require reopening it.

6. Worked example: a soil reference lot shifts one method

Back to contents

A laboratory replaces a sediment reference lot used for elemental analysis. Certificate values overlap within their stated uncertainties, but twelve paired runs show the new lot reads three percent higher on one digestion method and not on a second. Routine independent controls remain stable.

Review finds that the lots differ in particle-size distribution and the first digestion is more sensitive to incomplete homogenisation. The laboratory increases mixing and sample mass according to the new certificate, repeats the bridge and removes most of the shift. It documents a small residual uncertainty rather than applying a blanket three-percent correction to historical archaeological results.

First-principles check: State the decision this chapter protects, the evidence needed for it and the limit that remains. Write the rule before looking at the preferred outcome. Then identify the authority, observation or calculation that could change the answer. This keeps the method tied to evidence rather than hindsight.

Worked practice: Reconstruct the paired differences and decide which effect belongs in the method. Keep the intermediate steps visible. A reader should be able to locate the input, apply the same rule and understand why a different result would follow if one fact changed.

Changed-context transfer: The same shift appears on both digestion methods and two instruments. Preserve the underlying principle but replace the setting-specific details. State which parts travel, which do not, and which new authority or measurement would be needed.

Evidence record: Preserve raw runs, certificates, control results, investigation, corrective action, rerun and decision. Retain uncertainty and disagreement instead of compressing them into a clean label. The record should let another person reconstruct the decision without relying on the memory or status of the original decision-maker.

Decision-quality drill: Run the practice once under the most favourable plausible assumptions and again under the least favourable plausible assumptions. Compare which conclusion changes, which merely becomes less precise and which survives. If the result depends on an unstated convention, make that convention visible before acting. This is a transfer test: the method should guide a new case, not only explain the example after the answer is known.

Practical verification: Give the case file to a reviewer who did not build the analysis. Ask that person to reproduce the critical step, identify the stopping rule and name the strongest live alternative explanation. Resolve differences by returning to the governing evidence, not by averaging opinions. Record corrections, version the decision and keep a short list of conditions that would require reopening it.

7. Estimate the bridge and its uncertainty

Back to contents

Report paired differences, ratios or regression according to how results scale. Inspect heteroscedasticity and influential runs, and give confidence intervals or an uncertainty contribution rather than only a p-value. Equivalence requires a prespecified practically acceptable region; non-significance alone can reflect low power.

If a correction is adopted, define its domain and propagate uncertainty. A bridge factor estimated at high concentration may not apply near detection limits, and a single factor can be wrong when proportional and constant biases coexist.

First-principles check: State the decision this chapter protects, the evidence needed for it and the limit that remains. Write the rule before looking at the preferred outcome. Then identify the authority, observation or calculation that could change the answer. This keeps the method tied to evidence rather than hindsight.

Worked practice: Estimate a ratio bridge with uncertainty and test it against an acceptance interval. Keep the intermediate steps visible. A reader should be able to locate the input, apply the same rule and understand why a different result would follow if one fact changed.

Changed-context transfer: A regression reveals a nonzero intercept near the blank. Preserve the underlying principle but replace the setting-specific details. State which parts travel, which do not, and which new authority or measurement would be needed.

Evidence record: Keep data, exclusions, model, diagnostics, estimates, uncertainty and applicable range. Retain uncertainty and disagreement instead of compressing them into a clean label. The record should let another person reconstruct the decision without relying on the memory or status of the original decision-maker.

Decision-quality drill: Run the practice once under the most favourable plausible assumptions and again under the least favourable plausible assumptions. Compare which conclusion changes, which merely becomes less precise and which survives. If the result depends on an unstated convention, make that convention visible before acting. This is a transfer test: the method should guide a new case, not only explain the example after the answer is known.

Practical verification: Give the case file to a reviewer who did not build the analysis. Ask that person to reproduce the critical step, identify the stopping rule and name the strongest live alternative explanation. Resolve differences by returning to the governing evidence, not by averaging opinions. Record corrections, version the decision and keep a short list of conditions that would require reopening it.

8. Update control charts without erasing the transition

Back to contents

A new lot often requires new expected values or limits. Do not splice normalised results into one smooth chart without marking the transition. Preserve raw measurements and the assigned value used at the time, then show any standardised series with its transformation.

Estimate new limits from enough stable data; avoid widening limits merely to absorb a genuine method problem. During the transition, monitor old and new lots or an independent control so that a lot shift can be distinguished from concurrent analytical drift.

First-principles check: State the decision this chapter protects, the evidence needed for it and the limit that remains. Write the rule before looking at the preferred outcome. Then identify the authority, observation or calculation that could change the answer. This keeps the method tied to evidence rather than hindsight.

Worked practice: Redesign a control chart with a visible lot boundary and bridge annotation. Keep the intermediate steps visible. A reader should be able to locate the input, apply the same rule and understand why a different result would follow if one fact changed.

Changed-context transfer: The replacement lot arrives during a slow instrument trend. Preserve the underlying principle but replace the setting-specific details. State which parts travel, which do not, and which new authority or measurement would be needed.

Evidence record: Retain raw and transformed charts, limits, transition dates, bridge, control decisions and reviewers. Retain uncertainty and disagreement instead of compressing them into a clean label. The record should let another person reconstruct the decision without relying on the memory or status of the original decision-maker.

Decision-quality drill: Run the practice once under the most favourable plausible assumptions and again under the least favourable plausible assumptions. Compare which conclusion changes, which merely becomes less precise and which survives. If the result depends on an unstated convention, make that convention visible before acting. This is a transfer test: the method should guide a new case, not only explain the example after the answer is known.

Practical verification: Give the case file to a reviewer who did not build the analysis. Ask that person to reproduce the critical step, identify the stopping rule and name the strongest live alternative explanation. Resolve differences by returning to the governing evidence, not by averaging opinions. Record corrections, version the decision and keep a short list of conditions that would require reopening it.

9. Protect comparability in archaeological time series

Back to contents

Long-term archaeological series may combine results generated under several lots, methods and calibration eras. Harmonisation should retain the measurement actually reported and add a clearly labelled adjusted value only when the bridge supports it. Never overwrite the historical record.

Assess whether substantive conclusions change under raw, bridged and uncertainty-expanded analyses. When comparability is weak, stratifying by era or limiting a trend claim can be more honest than forcing a single continuous scale.

First-principles check: State the decision this chapter protects, the evidence needed for it and the limit that remains. Write the rule before looking at the preferred outcome. Then identify the authority, observation or calculation that could change the answer. This keeps the method tied to evidence rather than hindsight.

Worked practice: Reanalyse a ten-year site series across three lot eras. Keep the intermediate steps visible. A reader should be able to locate the input, apply the same rule and understand why a different result would follow if one fact changed.

Changed-context transfer: Only the middle transition has direct overlap data. Preserve the underlying principle but replace the setting-specific details. State which parts travel, which do not, and which new authority or measurement would be needed.

Evidence record: Keep original values, lot IDs, bridge versions, adjusted values, uncertainty, sensitivity and claim limits. Retain uncertainty and disagreement instead of compressing them into a clean label. The record should let another person reconstruct the decision without relying on the memory or status of the original decision-maker.

Decision-quality drill: Run the practice once under the most favourable plausible assumptions and again under the least favourable plausible assumptions. Compare which conclusion changes, which merely becomes less precise and which survives. If the result depends on an unstated convention, make that convention visible before acting. This is a transfer test: the method should guide a new case, not only explain the example after the answer is known.

Practical verification: Give the case file to a reviewer who did not build the analysis. Ask that person to reproduce the critical step, identify the stopping rule and name the strongest live alternative explanation. Resolve differences by returning to the governing evidence, not by averaging opinions. Record corrections, version the decision and keep a short list of conditions that would require reopening it.

10. Respond when acceptance fails

Back to contents

A failed bridge is information, not an instruction to repeat runs until the difference disappears. Verify transcription, storage, preparation, sample mass, instrument status and certificate interpretation. Consult the producer where appropriate and use independent materials to localise the cause.

Possible outcomes include method adjustment, new limits, a restricted use, a documented correction or rejection of the lot for the intended purpose. Quarantine affected results until the decision is made. Record negative findings and competing explanations.

First-principles check: State the decision this chapter protects, the evidence needed for it and the limit that remains. Write the rule before looking at the preferred outcome. Then identify the authority, observation or calculation that could change the answer. This keeps the method tied to evidence rather than hindsight.

Worked practice: Build a cause tree and stop rule after a bridge exceeds the allowable difference. Keep the intermediate steps visible. A reader should be able to locate the input, apply the same rule and understand why a different result would follow if one fact changed.

Changed-context transfer: The producer confirms a formulation change not reflected in the catalogue name. Preserve the underlying principle but replace the setting-specific details. State which parts travel, which do not, and which new authority or measurement would be needed.

Evidence record: Keep failure notice, quarantine, investigation, communications, tests, disposition and affected-result review. Retain uncertainty and disagreement instead of compressing them into a clean label. The record should let another person reconstruct the decision without relying on the memory or status of the original decision-maker.

Decision-quality drill: Run the practice once under the most favourable plausible assumptions and again under the least favourable plausible assumptions. Compare which conclusion changes, which merely becomes less precise and which survives. If the result depends on an unstated convention, make that convention visible before acting. This is a transfer test: the method should guide a new case, not only explain the example after the answer is known.

Practical verification: Give the case file to a reviewer who did not build the analysis. Ask that person to reproduce the critical step, identify the stopping rule and name the strongest live alternative explanation. Resolve differences by returning to the governing evidence, not by averaging opinions. Record corrections, version the decision and keep a short list of conditions that would require reopening it.

11. Audit the chain from certificate to published dataset

Back to contents

An audit should trace purchase and receipt, lot identity, storage, certificate, preparation, bridge design, raw measurements, analysis, acceptance and downstream transformations. Verify that the lot recorded in instruments and worksheets matches the physical container.

Sample published or archived archaeological results and reconstruct which lot and bridge applied. Look for silent default changes in laboratory software, reused spreadsheets and missing transition notes. A defensible bridge is useful only if it reaches the data products that depend on it.

First-principles check: State the decision this chapter protects, the evidence needed for it and the limit that remains. Write the rule before looking at the preferred outcome. Then identify the authority, observation or calculation that could change the answer. This keeps the method tied to evidence rather than hindsight.

Worked practice: Audit one lot transition from receiving log to a published table. Keep the intermediate steps visible. A reader should be able to locate the input, apply the same rule and understand why a different result would follow if one fact changed.

Changed-context transfer: A laboratory information system reused the old assigned value for six weeks. Preserve the underlying principle but replace the setting-specific details. State which parts travel, which do not, and which new authority or measurement would be needed.

Evidence record: Archive custody, certificates, worksheets, software settings, corrections, notifications and audit closure. Retain uncertainty and disagreement instead of compressing them into a clean label. The record should let another person reconstruct the decision without relying on the memory or status of the original decision-maker.

Decision-quality drill: Run the practice once under the most favourable plausible assumptions and again under the least favourable plausible assumptions. Compare which conclusion changes, which merely becomes less precise and which survives. If the result depends on an unstated convention, make that convention visible before acting. This is a transfer test: the method should guide a new case, not only explain the example after the answer is known.

Practical verification: Give the case file to a reviewer who did not build the analysis. Ask that person to reproduce the critical step, identify the stopping rule and name the strongest live alternative explanation. Resolve differences by returning to the governing evidence, not by averaging opinions. Record corrections, version the decision and keep a short list of conditions that would require reopening it.

12. Frequently asked questions

Back to contents

Can overlapping certificate uncertainty prove lots are interchangeable? No; intended use and laboratory behaviour also matter. Is a failed significance test evidence of equivalence? No. Must every lot change trigger a full validation? Not always; the scale should follow risk, material change and intended use.

Can old results be corrected? Only with a justified bridge, preserved originals and uncertainty. What if no old material remains? Use indirect evidence and state the weaker inference. Should control limits stay fixed? Only when their expected value and performance basis remain valid.

First-principles check: State the decision this chapter protects, the evidence needed for it and the limit that remains. Write the rule before looking at the preferred outcome. Then identify the authority, observation or calculation that could change the answer. This keeps the method tied to evidence rather than hindsight.

Worked practice: Prepare guidance for analysts receiving a replacement lot. Keep the intermediate steps visible. A reader should be able to locate the input, apply the same rule and understand why a different result would follow if one fact changed.

Changed-context transfer: Explain why a bridge is different from routine replicate precision. Preserve the underlying principle but replace the setting-specific details. State which parts travel, which do not, and which new authority or measurement would be needed.

Evidence record: Maintain reviewed answers, examples, source documents, escalation routes and revision date. Retain uncertainty and disagreement instead of compressing them into a clean label. The record should let another person reconstruct the decision without relying on the memory or status of the original decision-maker.

Decision-quality drill: Run the practice once under the most favourable plausible assumptions and again under the least favourable plausible assumptions. Compare which conclusion changes, which merely becomes less precise and which survives. If the result depends on an unstated convention, make that convention visible before acting. This is a transfer test: the method should guide a new case, not only explain the example after the answer is known.

Practical verification: Give the case file to a reviewer who did not build the analysis. Ask that person to reproduce the critical step, identify the stopping rule and name the strongest live alternative explanation. Resolve differences by returning to the governing evidence, not by averaging opinions. Record corrections, version the decision and keep a short list of conditions that would require reopening it.

13. The gold standard makes continuity testable

Back to contents

Long-term comparability is not created by a familiar label on a new bottle. It comes from a defined reference-material role, certificate review, planned overlap, representative matrices, blocked measurements and an uncertainty-aware acceptance rule.

Raw results remain traceable, transitions remain visible and downstream series are harmonised only within the evidence. When equivalence fails, the laboratory investigates and limits claims instead of smoothing the break. That discipline lets archaeological conclusions survive changes in materials without pretending the materials never changed.

First-principles check: State the decision this chapter protects, the evidence needed for it and the limit that remains. Write the rule before looking at the preferred outcome. Then identify the authority, observation or calculation that could change the answer. This keeps the method tied to evidence rather than hindsight.

Worked practice: Review a complete bridge and ask whether another laboratory could reproduce the decision. Keep the intermediate steps visible. A reader should be able to locate the input, apply the same rule and understand why a different result would follow if one fact changed.

Changed-context transfer: Repeat after the method itself changes at the same time as the lot. Preserve the underlying principle but replace the setting-specific details. State which parts travel, which do not, and which new authority or measurement would be needed.

Evidence record: Keep intended use, certificates, bridge protocol, data, decisions, controls, transformations and audits. Retain uncertainty and disagreement instead of compressing them into a clean label. The record should let another person reconstruct the decision without relying on the memory or status of the original decision-maker.

Decision-quality drill: Run the practice once under the most favourable plausible assumptions and again under the least favourable plausible assumptions. Compare which conclusion changes, which merely becomes less precise and which survives. If the result depends on an unstated convention, make that convention visible before acting. This is a transfer test: the method should guide a new case, not only explain the example after the answer is known.

Practical verification: Give the case file to a reviewer who did not build the analysis. Ask that person to reproduce the critical step, identify the stopping rule and name the strongest live alternative explanation. Resolve differences by returning to the governing evidence, not by averaging opinions. Record corrections, version the decision and keep a short list of conditions that would require reopening it.