Results
results - note these assume no measurment error as there is neglagable error in the computer
internal vs population strain differences for various correlation schemes (splitting test)
pick reasonable schemes - not 1 or -1 as unlikely to ocour in nature... other reasonings?
investigate repeatability of the splitting test - how well does each one predict the average of each tree which we know as it is an input.
same schemes - reliability of strain gauges vs number of gauges -- sugest an aproprate number of gauges assuming they have no measurment error
Compare above to argo paper and give crude estimate of measurment error under different schemes
Discussion
discussion
Conclusion
conclusion