The Holy Grail of genealogy is proof—of names, dates, relationships, or whatever else we hope to establish. But the best genealogies aren't necessarily those which assert their conclusions with certainty.
Certainty is an unattainable goal—one we should aim at, but not expect to meet. We can exhaust all available resources and still be wrong. Even a well-established "fact"—say, the birth date of Abraham Lincoln—may be called into question by a chance diary entry, or a newly discovered Bible record. The best we can hope for is a high degree of certainty—as high as the available evidence allows.
Degrees of certainty (or probability) should be assigned to any alleged fact. In my own research, I use a scale of descriptive terms:
"Possibly" (or "perhaps") — low probability (evidence is wanting)I also use two degrees of improbability: "probably not," and "certainly not." (These I often use when my research disproves an accepted opinion, to show that I have considered and rejected that opinion.) Certainty is more easily claimed when denying, than when affirming, an alleged fact: Abraham Lincoln was certainly not born in 1938.
"Probably" (or "presumably") — medium probability (evidence exists)
"Most Probably" — high probability (only conclusive evidence is wanting)
[No qualification] — near certainty (conclusive evidence exists)
Here's an example of how I would assign degrees of probability. John McDonald, aged 40 years, appears in the 1850 census with a 40-year-old woman and a 10-year-old boy, both bearing the McDonald surname. No relationships are noted. The man and woman, I might conclude, are probably (presumably) married, and the boy is probably their son. If the man and woman were fifteen years older, or fifteen years younger, the boy was possibly (perhaps) their son. If I find that the couple married in 1839, and that no other McDonald family lived in the vicinity, I might conclude that the boy was most probably their son. If the man and woman were twenty years older, or twenty years younger, the boy was probably not their son. If I find a birth record for the boy consistent with his age in 1850, I would declare that he was their son, without qualification. I could still be wrong, and further research would be warranted in any case.
How liberal one is with these labels may depend on one's gut reaction to the evidence, and dozens of external factors, including:
The area one is researching (Densely or sparsely populated?)Items marked as less probable should be targets for future research. As more and better data becomes available, "possible" facts can be upgraded to "probable"—and perhaps some unqualified facts will need to be downgraded in probability.
The mobility of the population (Horse-bound or train-riding?)
The rarity of the names involved (Was it the only Rabinowitz family then living in Kansas?)
The reliability of the source (17th-century church records or Joe Blow's WorldConnect page?)
Even if, after years of research, some dates or relationships remain only "probable," this is no cause for alarm, and no bar to publication. The genealogical community understands that those who jump to conclusions often fall short.