Archaeogenetics Without the Headlines

What survives in ancient remains

After death, DNA fragments and accumulates chemical damage while environmental microbes and modern handlers add contamination. Preservation varies with temperature, soil, skeletal element and treatment. A successful sample is therefore a selected biological archive, not a neutral cross-section of the past.

Authentication before ancestry

Researchers use dedicated clean laboratories, extraction blanks, fragment length, terminal cytosine-to-thymine damage, mitochondrial or sex-chromosome checks and replication patterns to distinguish ancient molecules from contamination. Statistical sophistication cannot rescue unauthenticated data.

From reads to genotypes

Low endogenous DNA often produces sparse coverage. Many studies call one observed allele at selected positions rather than a complete diploid genome. Capture panels improve comparability but condition analysis on chosen variants; shotgun sequencing samples more broadly but can be expensive.

Relatedness, sex and mobility

Genomes can estimate biological sex, close kinship and genetic similarity. Combined with isotopes and burial context, these findings illuminate residence and mobility. Biological relatedness does not establish household roles, inheritance rules or gender identity without other evidence.

Principal components and clustering

PCA and model-based clustering summarise genetic covariance relative to reference samples. Their axes and coloured components depend on datasets and modelling choices. A cluster is a statistical pattern, not an ancient tribe or self-identified people.

Admixture models are hypotheses

Statistics such as f-statistics and qpAdm can test whether a target is consistent with mixtures of reference-related ancestries. Source labels describe sampled proxies, often separated from the true source by time or distance. Several models may fit, and an accepted model is not unique proof.

Chronology changes interpretation

Direct radiocarbon dates and calibrated ranges locate individuals in time. Pooling samples across centuries can manufacture apparent populations or obscure change. Archaeological phases, reservoir effects and sample association must be reported explicitly.

Sampling bias is structural

Cool climates and petrous bones preserve DNA better than many tropical contexts; elite or excavated cemeteries are overrepresented; destructive sampling limits access. Ten individuals cannot automatically represent a city, language community or civilisation.

Genes do not encode language or culture

Languages and beliefs are learned and can spread with migration, elite dominance, marriage, trade or institutional change. Genetic movement can constrain historical scenarios, but mapping an ancestry component directly onto a language family or modern political identity exceeds the data.

Ethics and stewardship

Human remains are people, not only datasets. Good practice follows local law, plans research before sampling, minimises damage, involves custodians and affected communities, documents provenance and makes data reusable while addressing legitimate restrictions.

Reproducibility and uncertainty

Papers should release reads, alignments, sample metadata, dates, code and model specifications. Readers should inspect contamination, missingness, reference composition, standard errors, alternative models and whether conclusions survive removal of influential samples.

A disciplined historical synthesis

The strongest studies begin with an archaeological question, sample across relevant places and periods, and integrate genetics with material culture, isotopes, inscriptions and linguistics. They distinguish measured results from demographic models and broader historical interpretation.