Dorcas Hanna (Glasgow) Young

Automated leads use birth surnames—not married surnames—as parental eligibility, role-specific biological age limits, bounded and approximate-date uncertainty, household chronology, specific time-aware localities, mapped distance, tenancy succession, migration destinations and sex-aware Scottish naming positions across generations. People already attached as the subject's parents stay in the identity summary and are excluded from this alternative-candidate list, regardless of relationship confidence. The main calibration uses clues available on disconnected profiles; existing parent and sibling links are retained as comparison context rather than treated as transferable evidence. Common ordered names and grandparent-to-child echoes receive a small non-zero weight, so a John or James match can support several independent criteria without producing a lead alone. Rare names, repeated bearers and bundles across otherwise distinct branches receive substantially more weight according to catalogue frequency. An exact migration place is weighted by its rarity and repetition across both branches; a shared destination country is worth only one point. Extra weight is added only when at least three independent clue types converge: naming, locality, specific migration, household chronology or collateral networks. Sparse profiles retain a clearly labelled chronology-only fallback instead of disappearing. Countries, states, counties and research clusters do not count as a shared locality, and fallback, broad representative or kin-inferred coordinates are excluded from distance scoring. Shared descendant profiles are removed before branch-name comparison. A candidate already recorded as a parent-in-law is excluded. A confident or independently supported parent is treated as settled; uncertain and unassessed attachments remain open to alternative comparison. A different linked parent, an uncle/aunt or grandparent path, and a matching existing child profile prevent an alternative from being classified as a strong direct-parent lead until the underlying tree conflict or duplicate is resolved. Familiar forms such as Bessie are normalised. Surname remains an eligibility filter worth zero points.

Naming customs, shared farms and occupations are circumstantial. Children, spouses and collateral relatives may be missing, and tree-derived family networks are not independent evidence. Scores rank research leads; they do not prove parentage.

Dates are eligibility gates, not merely deductions: chronologically impossible people are excluded. Dates, places, spouses, parent certainty and children's profiles and names are then compared. A high score is a prompt to inspect records, not a merge recommendation.