The past is investigated through traces that exist now: light reaching a telescope, a material sample, a manuscript or a saved detector record. The central question is how each trace could have been produced and preserved.
A trace has more than one history
Consider a manuscript page. Its material has a history, the inscription has a history, and the surviving object may have passed through copying, repair, rearrangement and digitization. A date assigned to the material is not automatically the date of the composition, and a digital image's file date is not the date of either.
The same distinction applies to a laboratory record. The physical event, the acquisition, a corrected data file and the final plot may have different timestamps. A later copy can faithfully preserve an earlier record; its later creation date does not make its content a later measurement.
Reconstruction needs those relations rather than a single undifferentiated label called “the date.” When the relations are uncertain, that uncertainty belongs in the historical inference.
A physical clock needs a preparation model
For an ideal closed sample with a known initial population N₀ of a radioactive isotope, N(t) = N₀ exp(−λt). The elapsed time inferred from a surviving fraction f is t = −ln(f)/λ. At f = 1/4, this is two half-lives. That calculation depends on knowing the starting fraction and on the sample behaving as the assumed closed system.
Radiocarbon chronology requires more than applying one exponential to a measured number. The IntCal20 research supplies a calibration curve connecting Northern Hemisphere radiocarbon ages with calendar ages, accounting for variation in atmospheric radiocarbon through independently dated archives.
Contamination, reservoir effects and the relation between a dated sample and the event of interest can require additional treatment. Dating a piece of wood does not by itself date every later use of that wood. The physical measurement and the historical attribution must be joined by an argument about provenance.
Count independent traces, not copies
Suppose a historical hypothesis H begins with prior odds of one to four against an alternative. A trace has likelihood ratio three in favour of H. The posterior odds become three to four, corresponding to probability 3/7 within that specified two-hypothesis model.
If a second trace provides conditionally independent evidence with the same likelihood ratio, the odds become nine to four, or probability 9/13. But if it is merely a perfect copy of the first trace, multiplying by three again is unjustified. The copy can help establish transmission fidelity without supplying a second independent encounter with the original event.
This is an invented numerical illustration, not a probability assignment to an actual manuscript. Its point is that the number of surviving reports and the number of independent evidential paths are different quantities. Shared ancestry and shared errors matter.
Keep the passage and its interpretation connected
For textual research, a useful claim identifies the witness or edition, the passage locator, the reading adopted and any editorial intervention. A translation should preserve a route back to the source passage, especially where a temporal or metaphysical term admits several interpretations.
The essay on translating temporal concepts links the online books with Arabic source passages. Such links let readers examine the evidence and challenge an interpretation. Modernizing punctuation or resolving a word division may improve readability, but it should remain distinguishable from the witness itself.
A shared reading in several editions can reflect a shared predecessor. A disagreement can reveal an editorial choice, transmission history or a simple error. Neither agreement nor disagreement has a fixed meaning without that context.
The same discipline applies to remote physical events
An astronomical signal also requires a model of production, propagation and detection. Its reception is a present event carrying information about an earlier source. Inference may depend on cosmological parameters, foregrounds and instrument response, as discussed in light as a cosmic record.
For SMM/DTT, a present completed record does not uniquely specify every detail of its inner or outer history. The expanded book's equivalence of histories makes that limitation mathematically meaningful relative to an observation family. A new trace or intervention may separate histories that earlier records left indistinguishable.
A strong reconstruction therefore states both what the traces support and what remains underdetermined. Historical knowledge can be substantial without claiming access to every detail of the vanished event.
Sources and further reading
- P. J. Reimer and colleagues, The IntCal20 Northern Hemisphere radiocarbon age calibration curve (0–55 cal kBP), Radiocarbon 62 (2020), 725–757.
- W3C, PROV Data Model: explicit relations between records, activities and derivations.