Very interesting results. I suspect that humidity change was the reason. I have at least three humidity meter, in every room I store wood or other materials. It is easy to check relative humidity. For example, I store my linen threads in the basement, where the relative humidity is always about 50-70 %, depending of the season.
Bocken's breaking strength was 13,1 # (min. 11,4, max 15,5) with short sample (30 cm, n=17) and 10,8 (min. 9,7, max 12,2) with long sample (150 cm, n=12). Relative humidity 55 %, 20 C. So, if you use longish samples, we are getting just the same results.