Prepare for the Society of Genealogists Certificate in Genealogy (UK) by training three linked habits: (1) classify every source you use as original or derivative, and every piece of information as primary or secondary; (2) resolve identity questions with explicit evidence, not first-match searching; (3) write short proof summaries that cite what you viewed and address conflicts. For administrative details such as current structure, fees and entry requirements, check the Society of Genealogists directly.
Original versus derivative sources: why what you viewed changes what you can claim
An original is the record created at or near the event; a derivative is an index, transcript or abstract made from it. Separately, information is primary if supplied by an eyewitness. Classifying both, and citing what you actually viewed, is the foundational habit to build.
Keep the two classifications separate in your mind. Source level describes the artefact: the original marriage register, a scanned image of it, a transcribed index entry, or a family tree that cites nothing. Information quality describes the knowledge of the person supplying the detail: a widow reporting her late husband's birth date is a secondary informant even in an original source. A scanned image of a parish register is a faithful reproduction of an original; a published index is several interpretive steps removed.
Apply this by recording, in every citation, the form you consulted: 'Berkshire parish register, digital image' rather than a bare database reference. Then use the derivative's known weaknesses to decide when to go further. Indexers misread surnames, omit occupations, and flatten variant spellings; transcripts drop marginal notes. If a conclusion rests only on an index hit, treat it as a lead to the original, not as evidence in its own right.
Civil registration certificates: reading the informant, not just the facts
England and Wales civil registration began in 1837, and certificates record who supplied the information. Judge each field by that informant's knowledge, and expect delayed or late registrations to distort dates.
Worked scenario: a public online tree gives a man's birth as 3 March 1861 in Sheffield, citing nothing. The tempting move is to copy the date and move on. The better decision is to find the entry in the civil birth indexes, obtain or view the original certificate, and evaluate the informant: if the mother registered the birth herself, the date of birth is primary information from an eyewitness. If the father or another household member registered it weeks later, the date may still be accurate but deserves slightly more caution. The tree's date, by contrast, is an unsourced derivative of unknown origin and cannot settle anything.
Why it matters: a wrong birth date anchors a wrong identity, and every earlier generation you attach inherits the error. Make a habit of noting the registration district, the informant's name, description and residence, and the gap between the event date and the registration date. A long gap is a flag to check for a delayed registration or a second entry under a different spelling of the surname.
Census enumerations as a cumulative clue system, not a single proof
The censuses of 1841 to 1911 let you track households across decades, but each is a household snapshot reported by whoever answered the door. Use censuses to build corroborating patterns, then anchor identities on records closer to the events.
Learn the individual quirks rather than treating all schedules alike: the 1841 schedule rounds adult ages down to the nearest five years and records only whether the person was born in the county; later schedules give exact ages and fuller birthplaces; the 1911 census preserves the household schedule filled in by the householder, so you can see names crossed out and corrected. Each quirk tells you how much weight a particular age or birthplace can bear in an argument.
Turn this into practice: for one family, transcribe the relevant columns from every census between 1841 and 1911 into a grid, one row per person per decade. Expected observations: ages drift by a year or two between enumerations, birthplaces shift to the nearest town, and children appear and disappear as they leave home or die young. The grid teaches you which census details are stable enough to use for matching and which are only indicative.
The same-name problem: separating two candidates in one parish
Parish registers from 1538 onward contain many people sharing names and even parentage. Never accept the first baptism within a plausible date range; instead, build a candidate matrix that tests each candidate against independent evidence.
Worked scenario: a man's death certificate gives his age as 67 in 1889, pointing to a birth around 1821-1822 in a Norfolk parish. The register shows two John Climensions baptised there within eighteen months, sons of different fathers. The common mistake is to take the earlier entry and trace that father's line. The better decision is a candidate matrix: the marriage certificate names his father and the father's occupation; the census returns give a consistent birthplace and a matching occupation; the burial register's stated age narrows the window. Only the candidate whose father's name, occupation and household locations all fit survives the test.
Why it matters: choosing the wrong candidate silently substitutes an unrelated set of grandparents and every ancestor behind them, and the error is invisible to a reader who trusts your tree. The matrix also shows restraint: where both candidates remain plausible on the surviving evidence, the correct conclusion is 'unresolved', with the specific record that would settle it named, such as a will or a settlement examination.
Probate records: what wills and administrations actually prove, and what they omit
A will proves a testator's stated relationships and property at the date it was made; letters of administration often prove only death and next-of-kin. Neither proves an exact death date, and silence about a child is not proof of absence.
Read probate documents for their evidential structure. A will names beneficiaries and their relationships as the testator understood them, so it is strong evidence of kinship, but the information is only as accurate as the testator's knowledge and memory. Grants of administration name the administrator, usually a spouse or child, which helps identify next of kin. Probate registers and death records together can bracket a death between the last mention of a living person and the grant date, without you ever claiming a precise day the records do not state.
Avoid the two classic over-readings. First, a daughter described as 'my daughter' may be a stepdaughter or a child by a previous marriage; check the phrasing and corroborate with other records. Second, a child omitted from a will may have emigrated, already been provided for, or died, and the will alone cannot tell you which. Record probate findings as dated statements about relationships and property, then integrate them with parish and census evidence.
Citations and the proof summary: an exercise with a self-check rubric
Certificate-level work expects conclusions written as argued text, not just charts. The core exercise: take one ancestor, gather at least three independent sources, and write a short proof summary that classifies every source and addresses every conflict.
Set out a three-column evidence table before you write: source and form viewed (original, image, transcript), informant and whether the information is primary or secondary, and the specific claim it supports. Then write roughly 150 words concluding when and where your ancestor was born, citing each claim. If two sources disagree, say which you prefer and why, in terms of informant knowledge and source level, not habit or convenience. Naming what remains unknown is a strength of a proof summary, not a weakness.
Self-check rubric, scored out of ten with six as a reasonable milestone before mock practice: two points if every fact carries a citation stating the form you viewed; two points if each source and informant is classified; two points if conflicts are addressed explicitly; two points if the conclusion distinguishes what is proved from what is probable; two points if you named the one record that would most strengthen the case. Mark yourself honestly against the text, not the intention.
An adaptable preparation sequence and concrete readiness checks
A five-phase sequence works regardless of the exact assessment format: source inventory, classification drills, identity-resolution practice, paleography, then written proof summaries. Finish only when you pass concrete readiness checks against your own research, not when you feel familiar with the topics.
Phase one, inventory the major England and Wales record groups you can reach: civil registration, census, parish registers, probate, and the records of poor law and parish administration where available. Phase two, drill classification daily using your own finds. Phase three, resolve at least two same-name pairs with a candidate matrix. Phase four, practise transcribing early modern secretary hand and Latin-form parish entries from reproduced registers. Phase five, write a proof summary weekly and mark it with the rubric above.
Readiness checks: you can state, for any fact in your tree, the source form, informant and information type; you have one fully resolved and one honestly unresolved same-name case with written reasoning; you can transcribe a seventeenth- or eighteenth-century register entry with a date and named relationships; and you can produce a 150-word proof summary in one sitting. Administrative matters, including current eligibility, structure and fees for the qualification, are set by the Society of Genealogists and should be confirmed on its website.
| Record group | What it can establish | Typical informant | Watch for |
|---|---|---|---|
| Civil registration, England and Wales (from 1837) | Event dates, places, names, ages, occupations, father's name | A household member or official present at or near the event | Registration lag; informant guessing at dates they did not witness |
| Census, 1841-1911 | Household composition, ages, birthplaces, occupations across decades | Whoever answered the enumerator, or the householder in 1911 | Rounded ages in 1841; drifting ages and birthplaces; absent household members |
| Parish registers (from 1538) | Baptisms, marriages, burials before and alongside civil registration | Minister or clerk recording the event | Same-name candidates; gaps and losses; variant spellings; disputed entries |
| Probate: wills and administrations | Stated kinship, property, testamentary intentions; next of kin | The testator, or the administrator after death | Omitted children; step-relations; no exact death date in a grant itself |
References and further reading
Use these references to explore the concepts and check the latest information from the relevant organizations.