Editorial Commentary:

[Ed.] In 1990, at a session of the Joint Meetings of the American Mathematical Society and the Mathematics Association of America in Louisville KY, Professor Saunders Mac Lane served as keynote speaker for a 50th anniversary celebration of Mathematical Reviews.  In noting change over time within this publication, he cited two newer and perhaps exciting extensions of the Reviews: to Mathematical Geography and to Neural Nets.  Since then, MR has noted a variety of articles on these topics.  The next set of two essays represents some continuing attempts to combine these fields.  In forging such interdisciplinary links, it is helpful to note not only when tools from one field shed light on the other but also to note when they do not.  Such efforts may also suggest further direction for research for those interested in beginning to explore these topics. 

It is in the latter vein that specific commentary from John D. Nystuen is offered below (in reference to the first of the two essays).
Nystuen's challenges:

• Offer extended analysis:
• Use time-series data over a longer period
• Try a lagged version to consider periodicity in data
• Give meteorological reasons for expecting the regression to work or to fail
• Discuss underlying theory
• Justify, further, the purpose of the topic by discussing issues such as:
• At what amount of rainfall is the monsoon considered a failure for that year? 
  
• What is the critical minimum for crops that year?
• Mathematical observations suggesting further work (in reference to Figure 2):
• The predicted values seem to have an upward bias, rising to the top of the chart.  Why is that?
• The actual average appears to fluctuate around 210 mm.  One would expect the predicted avaerage to be near this figure.  Why is it consistently higher and is that difference critical?