Evaluating Evidence

by Dae Powell

It was summer 1997 when the Board of Certification of Genealogists rejected the phrase, "preponderance of evidence," or POE, as a standard of proof.

How do we prove a family relationship, if no record clearly contains the required information?   According to Elizabeth Shown Mills, author of Evidence! Citation and Analysis for the Family Historian (Baltimore: Genealogical Publishing Company, 1997), we must have three things:

  1. Good research
  2. Sound interpretation of evidence, and
  3. A carefully made argument.

When no record plainly states the data we need, we must resort to two basic strategies:

  1. Find reliable, alternative records that state the needed information, or
  2. Build a case from fragmented, circumstantial evidence.

Two topics affect how strong our case is, as follows:

  1. Our Construction Materials:  The quality of our evidence, and
  2. Our Methods of Construction:  The quality of our research and analysis.

And what if our evidence disagrees one with another? How do we build a solid and persuasive case for at least one set of data? Whom must we convince? The DAR, the next generation or even contemporary researchers, ourselves? What kind of evidence is convincing and how much of it do we need?

Evaluating Evidence:  Original and Derivative Sources

In a previous presentation I discussed Source Citations and Documentation. This brings us to "evaluating evidence." There are lengthy books written on this subject, but put briefly, evidence comes from two general sources: "Original" or "Derivative." The broad definition of an original source is as follows:

  1. A record made at or near the time of the actual event described and, preferably,
  2. Recorded by someone with firsthand knowledge of the event and
  3. Who had no vested interest in what the record says.

In short, a contemporary, unbiased record of an event. The "vested interest" part simply means that when you examine records, you need to be alert to the fact that people say or write things that tend to reflect their perception of those events in terms of their own self-interest. A simple example might be aunt Tillie's propensity for omitting a marriage or two of hers.

Original records include deeds, wills, court proceedings, church documents which record life events, e.g., births, baptisms, marriages, deaths or burials, birth and death certificates, and (usually) such documents as an application for a Social Security number (keeping in mind that the applicant may have had a reason for not telling the truth).

Derivative sources include almost all published works, be they county histories, genealogies or indexes and abstracts of original records. Please note that almost everything you will find on the Internet is a derivative source. In much of the US, the main repository of original records is generally the county in which the family lived (which is why it is important to record the county of residence when making entries into your database). The exception to this is that in most New England states, vital records were most often recorded at the town level. In other places, this may vary greatly.  For example, in Great Britain, Scottish birth, marriage and death records are to be found in Edinburgh — those for England and Wales in London. Other countries may categorize by the relevant state or province, or even hold everything centrally.

Some Original Records

The following, incomplete list deals with the strengths and possible pitfalls of various types of records.

Birth Certificates

These are generally reliable—keeping in mind that while they are never proof positive of who the father was, but they will be in many cases the only official statement of such. In more modern cases, it is perfectly legal to alter birth records to reflect such things as adoption. (The foregoing applies primarily to the US. It's extremely difficult, if not impossible, to change a certificate in Britain, where finding an original birth certificate can be the first step for adoptees wanting to learn about their biological parents. Generally, even on altered certificates, the date of birth is correct — although my cousin's birth is recorded in the court house as being 10 days after it actually was. (In any case, the exact date of birth is not usually a critical factor in establishing a genealogical link.)

"Delayed" birth certificates are an official substitute for a original record; often there will be some sort of back-up documentation on file at the court house to verify the date of birth. You need to know the local requirements for issuance of delayed certificates to know where to locate this backup info (sometimes just an affidavit is needed, but often an official court proceeding is involved). The availability of birth records varies between jurisdictions; generally laws providing that these records be kept did not "take hold" until early this century. Keep in mind, too, that in the colonies of New England, births were recorded in the towns, while in colonial Virginia and elsewhere in the South, these records were maintained at the church parish level.

Marriage Records

Marriage bonds were prevalent until the latter part of the 19th century and usually contain only the name of the bride or groom. If the father or parents of either are given, this is an indication that the person was not yet 21 years old and also that there may be a parent's consent on file. Note that a marriage bond was not a "promise to marry," but the equivalent of a statement that there was no lawful reason that the marriage could not take place, such as either party not yet of age, either party already married or the parties were too closely related as defined by laws of the jurisdiction.

Keep in mind the distinction between the date the marriage bond was signed as opposed to the actual date of the marriage (usually only a few days later). If a minister's or J.P.'s name is given, this usually indicates the date is that of the marriage and is a "marriage return" that was required of clergy or justices. In the US marriage licenses generally replaced bonds about the last decade or so of the 1800s and usually provide more detailed information, such as age or date of birth of the parties and names of parents.

"Vested interest" can play a part here: Often a groom was a few months short of his 21st birthday when he married his bride and "fibbed" about it on the application. Here is a clue: If a marriage took place in a county other than where the couple lived, there may have been a reason. Thus, many couples were married in an adjoining county to avoid detection of their actual age; had they applied for a license in their home county, the clerk would have checked their birth records.

Keep in mind that the information on parents and dates of birth in marriage records are considered "derivative" sources: They do not reflect a contemporary event.

Death Certificates

These are original records only with respect to the date and place of death and cause of death (if signed by a doctor, coroner or medical examiner). Information regarding the date and place of birth of the deceased person or the names of his or her parents is derivative information and only as reliable as the "informant" named on the certificate.

Note that even if the informant can be considered a "reliable source," he or she may have been confused in responding to the questions, did not know the correct information or had a momentary lapse of memory with respect to a certain item. Example: The death certificate of one of my great grandmothers contains personal information about her provided by one of her sons. In response to the question: "Mother's maiden name?" he provided his mother's maiden name (i.e., the name of the deceased), rather than the name of her mother. Again, the information to be found can vary significantly, depending on the state or country you are researching. Using a British example again, certificates for births, marriages and deaths began in 1837 in England and Wales—but not until 1855 in Scotland.

Wills and Probate Records

These are usually the best evidence of descent. Keep in mind, however, that all children, living or dead, may not be mentioned. One man's will, for instance, mentions only the five children by his third wife and her two children by a previous marriage. Not mentioned were the 11 children by his first two wives.

Sometimes other heirs will be mentioned in other records relating to the probate. In some states (Ohio is one), the law required that a list of all heirs be recorded. However, this apparently was not always done and even if it was, the records were not well maintained; most court house workers are unaware of "heirship" books.

Deeds and Land Records

Deeds are often overlooked as an excellent source of relationships, particularly if the land was inherited by one or more children. When inherited land is sold, it is often necessary to state from whom it was inherited and the relationship to that person. If there is ambiguity or seemingly an error in a deed, check the previous and subsequent transactions involving the land to see if errors can be reconciled.

Guardianship and Administrator's Bonds

These were usually kept in separate books and, even if not indexed, are more easily searched than some county records. A good source not only for the names of minor children but the date of the death of an ancestor. And often those who give bond for these "jobs" may be relatives. Note that the purpose of appointing guardians for minor children was to protect their interests in property they are to inherit at their majority, not to be responsible for raising them (although this could also be true).

Often guardians for children were not appointed at the time a parent was lost, but at the time of some other event. Thus no guardian might be appointed until the widow planned to remarry or until the death of a parent of the children's father or mother — at which time they became heirs either by will or by law. The date of a guardian bond is a good clue for discovering, for example, the name of a grandparent: The death of a person with the appropriate surname could signify a relationship because it triggers a guardianship.

Other Court Records

Since general court records are often not indexed (as probate records, marriage records and deeds are), these records are often overlooked. Considerable individual research is often required but the rewards can be well worth it. Family spats, "petitions for partition" of land inherited by many heirs, paternity or support suits — or any other type of suit — can be extremely revealing. Be aware that some church records can have a status half way between original and derivative. While most clergy recorded baptisms, etc., at the time of the event, some were prone to procrastinate and to deciding for themselves how names should be spelled, rather than consulting their parishioners.

There are also cases where whole families were baptized en masse—and then the ages or birthdates of older children should be treated with caution. Similarly, in the German Lutheran and Reformed churches in the US colonies and even later, often the task of making entries in the church records fell to the father of each family because there was no regular minister to do the job. When the family joined the church, the father would enter the names and births of the older children, then add each new child on the page reserved for the family. In New England, when a family moved to another town, the clerk in the new town would sometimes enter the names and births of the entire family, even though these people were born elsewhere.

Some Derivative Records

Family and County published works vary in their accuracy and evaluation of the information within them must be based on their reputation and your experience in using them. While it is sometimes necessary to rely on these records, whenever possible they should be considered only as good clues for further research into the original records, especially with regard to the original lines you are researching.

Lesley Robertson notes that some printed books should be taken with no more than "a pinch of salt" — including many that claim to show connections to noble or royal houses. As an example, there was a limited edition of a book by George Parker Knowles with this grand title: A Genealogical and Heraldic Account of the Coultarts of Coulthart and Collyn... To Which Are Added the Pedigrees of Seven Other Considerable Families, etc., With a Genealogical Account of the Rosses of Daltons in the County of Dumfries. It has coats of arms, a large pedigree chart and is printed on vellum.

How much would you pay for such a wonderful tome?

Now consider what antiquarian bookseller Benny Gillies says about it in his latest catalogue. "An elaborate and expensive hoax," he writes. "The seven of the considerable families never existed. The man who had it published was of uncertain origin. The man who is believed to have been his grandfather was a half-witted small farmer known locally in Kells as 'Laird Cowtart.' The place and castle of Coulthart never existed and the arms were borrowed from the Essex family of Colt, and others."

While US County Histories can be notoriously unreliable, generally I consider them to be fairly accurate as to the birth dates of the person whose biography is being recorded and his or her immediate family. Too, they are generally accurate in stating relationships to persons contemporary to the publication date. But too often they are inaccurate in such details as state or country of origin of grandparents or great grandparents. Sometimes even the names of those in earlier generations get garbled, as do other relationships in earlier generations. A most common problem is that connections to famous people of that same surname are often postulated, or the patriotism, heroism or military service of earlier generations is, well, somewhat exaggerated.

Often the only information you will find comes from census records, a family Bible or a gravestone inscription. None of these can be considered entirely reliable.  Nor are they generally "original" sources.  Census records are both original and derivative sources. They are original in the sense that they are a (more-or-less accurate) enumeration of the population of the years in which they were taken and are a record of where a particular family lived that year (in and of itself valuable information). They are derivative in the sense that the information recorded about individuals is not a contemporary record of, for instance, the birth of any person. The accuracy of this data depends, first, on an informant providing correct information—and not even mothers remember the ages of all their children at every given moment. Fathers even call their children by the wrong names.  After being given accurate information, the enumerator must then have accurately recorded it and, since he often copied the day's work onto new forms — and then made the required additional copies for various jurisdictions — he must have copied it accurately. However, even if the specifics are off, census records usually are pretty good indicators of relationships and, at least, relative ages. And you often have the opportunity to verify your observations against a later or previous census.

Again, if your citation is solely to a census record or several censuses, the reader can make a judgment on the information based on that knowledge. In estimating year of birth from age in a census record, keep in mind the "official" date of the census. In the latter part of the 1800s, this date was June 1. Ages were supposed to be recorded as of that date, even though it might be September before the enumerator finished his rounds. Thus the year of birth for a person who is age 10 at June 1, 1850 has a 5/12 chance of being 1840 and a 7/12 chance of being 1839.

Bible records or gravestone inscriptions are usually considered to be derivative sources, but they are much better than Grandma's memory. The memories of grandparents and great aunts or uncles can be quite selective and the more distant the event in time, the less accurate the report on it is likely to be.

In evaluating a Bible record, you should first note the date the Bible was published. If it was published in 1859 and contains the birth record of someone born in 1801, obviously the record was made long after the fact and must be considered in that light. An important consideration is the handwriting in which the various entries were made.

Sometimes you can easily determine that all of the entries were apparently made by the same person at one time. In this case, you can surmise they were prompted by the last event so recorded. Bible records can generally be considered more accurate if the entries appear to have been made by the same person over a period of time. Look for such things as each entry being only slightly different; perhaps a more shaky hand is evident or a different writing instrument used. With this type of pattern it is likely that the entries were made at or near the time of each event recorded. It is common to see a series of entries in a Bible in which all appear to be different except perhaps two or three. In this case, what it usually means is that Mother or Father neglected to record the births of John and Alfred until William was born, at which time he or she recorded all three. Thus William's entry can be considered to be the most accurate and the other two subject to the vagaries of memory.

If two or three distinctly different hands are apparent in a Bible, it generally means that the task of maintaining the records was passed to the next generation at the death of the previous recorder. If Mother's death is recorded in a new hand and subsequent entries made in that hand, the conclusion you may reach is that she was the original record keeper and a son or daughter, or even a second wife, took over when the first wife died. You should also be aware that Bible records are a place where deliberate deception may be practiced. I know of a Bible record of a Kentucky couple which contains, among other entries, the marriage date of the couple and the birth date of their only child. The birth record follows the marriage date by a little more than a year. Across the Ohio River in an Indiana County is recorded the marriage of this couple on the same date but a year later than given in the Bible. Also recorded in this Indiana County is the birth of their daughter on the date given in the Bible, but only two months after the date the marriage was recorded.

Gravestones, unfortunately, can be erected long after a person dies or can be carved inaccurately. Generally, though, tombstones are more accurate with respect to date of death than the date of birth. Once when talking about genealogy to a fifth grade class, an alert lad expressed this well. After discussing original and derivative records, I asked if a tombstone inscription was original or derivative. "Derivative!" he declared.  "No one has his tombstone carved when he is born."

I want to emphasize as strongly as I can that most of the genealogical information you may discover on-line is strictly and totally derivative. It provides you with little more than clues to be checked out. This caveat applies not only — and especially — to website genealogies, but to the digitized versions of previously published works or transcriptions of local records. And to indexes — up to and including the International Genealogical Index (IGI), the assorted census indexes and anything requiring the intervention of a human for interpreting or recording.

An example: a while ago I encountered at a county website some inscriptions for a cemetery where one my great grandfathers and his father are buried. In the case of the former, a typo had been created in transcribing the material from the book in which it originally appeared. In the case of the latter, a copying or transcribing mistake in the original book was repeated. Two relevant records, both wrong.

The answer to the question, "Where on the Internet can I find my ancestry?" is nearly always: "You can't." What you can find, however, are the means to greatly reduce the time and effort needed to find the records for accomplishing your task. Back to the IGI, certainly one of the better digitized databases. I have encountered an entire Powell family—father, mother and some eight children—that is an amalgamation of several other families — but, as the family described, never existed. The Ancestral File and the various World Family Tree disks, so far as my experience with them, generally lack the thoroughness one should expect and usually cannot be relied on for anything but pointers to source material.


In a perfect genealogical world there would be a precise original record for every event in your database (and you, of course, would have recorded it). Unfortunately, the world of genealogy is one of the least perfect. Therefore, you will often have to rely on derivative sources to document your work. As to whether you actually need the original record in all cases, I think some common sense has to prevail. It is, after all, an all-but-impossible task for you to personally verify each item of information in a compilation that may contain the lives of a thousand different people.

Suppose, for example, you have a book of will abstracts for Jefferson County. Is it enough to use the information in the abstracts as documentation? Or do you need to get a copy of the actual will and personally verify that the abstract is correct? I know "sticklers" who will say you should not stop until you see the original record. I view this more practically and believe you can make a judgment on whether you should examine the original document based on context and the relative importance of this fact to the total work. If there seems to be a conflict as to whether the information in the abstract conforms to your previous knowledge of the "facts," by all means the original should be sought.

If all you are doing is giving the names of the parents of a spouse to a sibling of your great grandfather, I consider it enough to cite a published work. You might even cite the word of a descendant or other relative of the couple in such cases. If your citation is to a published work rather than to the original record (or a photocopy or transcription made by another researcher for you), your reader can evaluate the relative worth of the information. The same thing applies if your citation is to correspondence or a group sheet sent you by another person. If it turns out this is a line a reader may be particularly interested in, he or she has pointers to the records and can do the follow-up research.

One of the most important skills you can acquire will be learning to evaluate the accuracy of original records and the relative reliability of derivative sources, both with respect to specific records and in general.

Links about how to find original sources in various countries:
The following list, put together by Lesley Robertson, includes just a few examples; those of you in other areas of the globe should be able to locate similar sites:

Home Page Home Page, www.ShoeStringGenealogy.com