We can also look at how water activity itself behaves over time.
Here we see that there is a tendency for coffees to condense between around 0.525 and 0.5, given enough time.
This specific range is likely a function of our warehouse conditions more than anything intrinsic to coffee in general.
Averages over large sample sets can be deceiving.
Given the above chart, we might conclude that only coffees that PSS above Aw 0.610 (H) need be of concern.
The standard deviation of change in water activity (o of △) is quite high in general.
If our 1 standard deviation range for Aw in green coffee is from 0.497 Aw to 0.611 Aw, we have a primary working range of 0.114 Aw.
The average standard deviation of change (o of △) from PSS to ARR is 0.0, or about 46 percent of the primary range of water activity values that we're working with.
When we look at a statistic like average change we want to check and see if it is composed of many uniform changes or if it is composed of a more diverse collection of changes that cancel each other out.
Even though the mean outcome may be the same in both scenarios, the specific volatility may yield insights that a simple mean can miss.
Estimates of ARR Aw based on average change and outside average change (one standard deviation) show that the water activity measurements that we're looking at are very changeable.
These estimates for PSS arrival can be checked against the actual values shown for Round 2 in the bottom half of the table (highlighted).
The L category PSS average Aw is skewed low due to a handful of very low Aw samples.
It is notable that the L arrival Aw level is higher than the M1.
One reason that this is important is that the ill effects that we observe in green coffee that are due to water activity are not reversible.
Once lipids have oxidized, they do not un-oxidize just by bringing water activity back into an ideal zone.
Molds may be re-inhibited by reductions in water activity, but they are neither killed nor removed by Aw reduction.
Just because a PSS had water activity in an ideal zone does not mean that the coffee the sample represents will remain there.
A second reason that this is important is that it undermines the value of a single point water activity measurement taken of a sample sent to Minneapolis from Colombia, for example.
In the long run, we've seen that lower Aw coffees have performed better in some ways and to a certain degree, but this does not mean that any particular low-Aw pre-shipment coffee will perform as modeled.
출처 : CAFE IMPORTS
