Dr Isobel Clark FSAIMM FIMMM CEng
Thoughts from the author, 6th January 2015
Not so nice to know some things have not changed in 40 years: a very good friend bullied me into submitting a paper for the (now 3 volume set) memorial publication in honour of Danie Krige. Hard to believe it is almost 2 years since he passed on. After a lot of wrestling with my conscience, I put together a paper tracing "regression" and "kriging efficiency" from his earliest work in 1951 to the definitive paper in 1996 -- with a slight detour through my own formalisation of "geostatistical regression" published in 1983 (Math. Geol.). Submitted in October 2013, the reviewer's comments finally came back a year later rejecting the paper for publication. One of the comments was that I had failed to explain why this paper was important to the "science of geostatistics". Oh, my, the deja vu! That is what one of the referees for my 1974 paper said -- if not in those exact words! If you have not read it yet, there is no geostatistics in my 1974 paper (see comments below written over 10 years ago!!).
So the paper may not appear in the SAIMM Memorial Volume 3 and may also not be presented at the conference in Danie's honour this coming August. However, I have left my draft up for anyone to download, should you wish. Maybe, it is simply because I am old but it is my feeling that a historical perspective (a) prevents you from re-inventing the wheel and (b) diminishes that arrogance which comes with thinking you know it all.
Thoughts from the author, 28th February 2006
I was thinking that I should update this page as I update the links to downloadable papers, but I just read this through again. It sounds fine to me. One diagnostic of passing time, though. My son was born one month after the presentation of that 1974 paper. In December 2005, his first child was born. At 31 he is below the current average age for a first child. Nice to know some things have changed���.
Thoughts from the author, 28th May 2003
I have been scanning my 1974 paper on mixtures of populations, written jointly with Richard Garnett. Richard and I had found that teaming a numerate (then) statistician together with an experienced geologist produced amazing results � greater than the sum of the parts, indeed! This was my first paper published in a real, refereed journal (Trans Inst Min Metall) and we were asked to present it at a meeting of the Institution, together with papers by Andre Journel and Massimo Guarascio.
I remember the presentation well, being the only woman present and 8 months pregnant at that��������������
The discussion was published after this paper and two others were presented at a meeting of the I.M.M. in London in October 1974. A summary of the presentation is included in the published proceedings. I have removed sections which pertained only to the other papers in the session. However, some of the comments refer to all three papers, and these have been included below.
Re-reading the discussion, I am mildly depressed by how little things have changed in 30 years. Do people really think that we would use such complicated statistical methods if we could just go back and look at the core to see whether it was oxide or sulphide? Well, maybe they would, I wouldn't! You can also see clearly the axes ready to be ground. This paper was completely non-geostatistical, using the spatial layout of the samples only to identify the geological environment. However, we were criticized for not (somehow?) using semi-variograms to identify changes in geology. Given that semi-variograms assume complete homogeneity of the geology, how could we expect them to rear up in dismay and scream "no, no, there are mixtures of mineralisations here!!".
The other interesting thing about reading so far back is seeing the pre-cursors to future work. We used the same techniques recently in valuing the Skorpion deposit in Namibia -- because the multiple mineralization phases had not be recognized when the core was logged and no visual signs were present. The core had been logged by lithology, weathering and geotechnical parameters. However, none of these were related to the actual deposition mechanism and no geochemistry or mineralogical analysis had been carried out on what seemed (at that time) to be a simple deposit. That problem was discussed in my 2000 paper at MRE21 in Cardiff
I also notice that my aversion to the blanket use of accumulations was already apparent in this, my first real published paper! And my liking for the simplicity of the lognormal.
Finally, Hester's statement that "It has been used exclusively by the various authors in discussions on the gold deposits of South Africa (Sichel, Krige and others) whose works in this field are well known." beautifully illustrates an attitude I have fought all my professional career. Just because it has been done before, and for a long time, does not mean that there are not other ways to tackle the same problem.
It is our purpose to push forward the bounds of human knowledge, not to sit back and say "ah, well, someone else did this already, so I'll just do that then.". It may be that the 'old' way is the best. It may be that 'present' ways are the optimum. Or it may be that out there (somewhere) someone is inventing a whole new non-geostatistical way of tackling these problems which I will embrace with enthusiasm -- just as soon as I understand it!!
Share and enjoy.