I've been playing around with tracing a peripheral line on my family tree - the Gates family of East Haddam. They're cousins of an ancestor's half-siblings. Since they're not directly part of my tree, I feel less inclined to guarantee that my work is perfect. In short, they're good research practice.
Unfortunately, that doesn't mean the family is easy to trace. I found their birth information fairly quickly. Thanks to my own family records, I knew where they had lived as children. Using that information, I turned them up quickly on the 1860-1880 census enumeration. After that, things quickly dead-ended. Why? Because I was searching on the mother's name and, although I didn't know it at the time, she had died in 1892 at age 60. I might have accomplished my goal by switching to a child - although several died young and may not have shown up later. Something else, however, came to my rescue.
I remembered some earlier research I had done in East Haddam. The Gates family has been in the area since the 17th century. By the 18th century, they had begun to migrate but never went far. Most of the family could be found in either East Haddam or the northern part of Lyme. Assuming that the family followed their traditional pattern, my Gates should be found in East Haddam or Lyme. Bingo!
Lesson learned: with an older Connecticut family, make sure you know the surname migration patterns. They tend to repeat and learning them will save you some time.