I think I found the papers you are talking about in reference to CAG number in the huntington gene, specifically, the 36 CAGs - your memory is vindicated! Things start getting really strange in the huntington gene when it gets to 36 CAGs. I put statement from OMIM at the bottom of this response and you can skip over the other stuff - I put that in while I was trying to find the succinct answer. The OMIM statement is why I used 40 CAGs in my example, because incomplete penetrance can occur in people up to 40 CAGs. I did make a mistake in my example in the number of CAGs in normal alleles, normals are a little higher than I showed, but I was guessing. The OMIM statement also indicates that 1% of normal Causcasion population has an intermediate number of CAGs.
Here's an abstract by Duyao et al 1993.
http://www.ncbi.nlm.nih.gov/sites/entre ... ds=8401587
They give the range of 37 to 86 repeats in HD families, and 11-34 in normal chromosomes.
Here's another by Rubinsztein et al. 1996, in which there is people in the CAG length of 36-39 that made it into their 90s and still appeared unaffected.
http://www.ncbi.nlm.nih.gov/sites/entre ... ds=8659522
Here's an interesting one by Laccone and Christian 2000 in which the mom had 36 CAGs and was unaffected, but she had two daughters, one with 66 CAGs, and the other with 57 CAGs, and they both had HD in showing up in the second and third decade respectively. It is an example of CAG expansion in the maternal line; there are numerous papers on paternal expansions.
http://www.ncbi.nlm.nih.gov/sites/entre ... s=10712225
Here's one by Leeflang et al. 1995 that shows as the number of repeats in the high normal range increases, the probabiliy of further mutations in sperm that brings repeats into HD range goes way up. Someone with 36 CAGs put out 8% of gametes with at or more than 38 CAGs.
http://www.ncbi.nlm.nih.gov/sites/entre ... ds=8541834
I copied the statement below from the OMIM website on HD, which is a huge summary with links to all the papers.
"As outlined earlier, all mutations for Huntington disease (amplification of the CAG trinucleotide repeat in the coding region of the gene) arise from so-called intermediate alleles (IAs) containing between 29 and 35 CAG repeats. The CAG repeats expand on transmission through the paternal germline to 36 or more repeats. Intermediate alleles are present on approximately 1% of normal chromosomes of Caucasian descent. Affected individuals have an expanded allele of between 36 to 121 CAGs but incomplete penetrance has been found for repeat lengths of 36 to 40 CAGs. Using single sperm analysis, Chong et al. (1997) assessed CAG mutation frequencies of 4 IAs in families with sporadic HD and IAs ascertained from the general population by analyzing 1161 single sperm from 3 persons. They showed that the intermediate alleles of the former group were more unstable than those in the general population with identical size and sequence. Furthermore, comparison of different sized IAs and IAs with different sequences between the CAG and the adjacent CCG tracts indicated that DNA sequence is a major influence on CAG stability. These studies provided estimates of the likelihood of expansion to 36 or more CAG repeats for individuals in the 2 groups. For an IA with (CAG)35 in the family with sporadic HD, the likelihood for sibs to inherit a recurrent mutation equal to or more than (CAG)36 was approximately 10%. For intermediate alleles of a similar size in the general population, the risk of inheriting an expanded allele of 36 or more CAGs through the paternal germline was approximately 6%."