Regards, Jack Herz

Sample from one volume:

Volume 3 also contains an article on “Grading Systems of Scientific
Workers” by D. Rougge. Several systems are evaluated:
1. Idea grading: based on IPM (ideas per minute). It works quite well
for scientists employing many graduate or postgrad students.
2. Execution grading: based on projects – no matter whether original or
not. The important factor is the number of publications or patents.
3. Disagreeable grading : Certain activities (administration of a
department, planning of laboratories, organization of conferences,
inspection of laboratories) seem to be quite disagreeable to scientists.
Promotions can be keyed to the performance of these disagreeable jobs.
4. Public relations: Grading depends on scientists ability to convince
potential investors to part with their money

A verification algorithm that obviously eliminated sites with a cooling bias would look pretty suspect. Let’s hope that’s not the case.

The “>40% complete” threshold for AWS selection seems to be defined in terms of when the READER data record starts for each station. It would have made more sense to pick one common start time to define data completeness.

Table S1. Verification statistics (mean correlation coefficients (r), reduction of error (RE) and coefficient of efficiency (CE) statistics, from pre and post-1995 calibration/verification calculations) for the AWSbased reconstruction, for those 26 automatic weather stations where sufficient calibration data exist (data >40% complete) and verification skill is acceptable at >95% confidence.

“Verification skill” may reject negative trends.

Also there’s an ongoing loose end with Table S1 and S2 that I don’t understand. Harry and 3 other AWS stations are in Table S2 with the surface stations and not in Table S1 with the AWS stations. I can’t make head nor tail of the various “explanations” at RC on this.

It looks like AWSs Mount Siple and Elizabeth, both on West Antarctica and having negative trends, were tossed out before the “reconstruction”. If those had been included, I wonder how well the AWS reconstruction would have correlated with the model.