The stated aims of the Lecture Notes in Biomathematics allow for work
that is "unfinished or tentative". This volume is offered in that
spirit. The problem addressed is one of the classics of statistical
ecology, the estimation of mortality rates from stage-frequency data,
but in tackling it we found ourselves making use of ideas and techniques
very different from those we expected to use, and in which we had no
previous experience. Specifically we drifted towards consideration of
some rather specific curve and surface fitting and smoothing techniques.
We think we have made some progress (otherwise why publish?), but are
acutely aware of the conceptual and statistical clumsiness of parts of
the work. Readers with sufficient expertise to be offended should regard
the monograph as a challenge to do better. The central theme in this
book is a somewhat complex algorithm for mortality estimation (detailed
at the end of Chapter 4). Because of its complexity, the job of
implementing the method is intimidating. Any reader interested in using
the methods may obtain copies of our code as follows: Intelligible
Structured Code 1. Hutchinson and deHoog's algorithm for fitting
smoothing splines by cross validation 2. Cubic covariant
area-approximating splines 3. Cubic interpolating splines 4. Cubic area
matching splines 5. Hyman's algorithm for monotonic interpolation based
on cubic splines. Prototype User-Hostile Code 6. Positive constrained
interpolation 7. Positive constrained area matching 8. The "full method"
from chapter 4 9. The "simpler" method from chapter 4.
