Theory of Sampling and Sampling Practice, Third Edition

Francis F. Pitard

• Publisher: Chapman and Hall/CRC
• Year: 2019
• Print ISBN: 9781032032085
• eBook ISBN: 9781351105910
• Topic: Business
• Subtopic: Probability & Statistics
• Index: Business

A step-by-step guide for anyone challenged by the many subtleties of sampling particulate materials. The only comprehensive document merging the famous works of P. Gy, I. Visman, and C.O. Ingamells into a single theory in a logical way - the most advanced book on sampling that can be used by all sampling practitioners around the world.

Part I

Introduction and a Management Strategy

Introduction

Numerous articles and a few books have been written about sampling of particulate materials before this third edition. Then, why select Pierre Gy’s theory, Visman and Ingamells’s works? As a Pierre Gy’s Gold Medalist I want to bring my knowledge and experience on the Theory of Sampling (TOS) and contribute to making sure TOS grows in a rational way, in spite of its many detractors. Looking at comments made around the world, it is clear that many statisticians and empiricists promoting “Measurement of Uncertainty” (MU) strongly believe that TOS is something they can live without. Such antagonism is misplaced, unjustified and very unfair. I also strongly believe the MU promoters, who most of the time, are more comfortable with J. Visman’s work, need TOS, and vice versa. In this third edition of my book a special effort is made to integrate J. Visman and C. O. Ingamells’s works into the TOS and create a unified foundation that may help to create better sampling standards.

I have been closely associated with Pierre Gy’s work since 1969. Gy’s books are meant for university professors and sampling practitioners who should read them thoroughly, which is rarely the case; they read these books in what I call “in diagonal” missing many critically important subtleties. For those who are in charge of solving sampling problems every day, finding the adequate answer to their questions in these books may prove difficult, which is why they give up and look for more pragmatic empirical approaches. There was therefore an obvious need for a “digest” of Pierre Gy’s books and a long overdue attempt to unify his work with J. Visman and C. O. Ingamells’s works.

I have been also closely associated with C. O. Ingamells’s work (J. Visman’s school of thought) since 1973. A large amount of that time was spent in appreciating the respective strengths and weaknesses of these authors. Meanwhile, I discovered to my great surprise the inability of engineers and even scientists to deal effectively and rationally with sampling problems in general. It became natural for us, in close association with the founders of the World Conference on Sampling and Blending (WCSB), to believe that an aggressive education program was needed worldwide.

My association with Pierre Gy, C. O. Ingamells, and WCSB members provided me with the fundamental basis of an elaborate educational program and quickly confirmed what I already knew for some time that the work of Dr. Pierre Gy on sampling of particulate materials is unique, and statistically correct. However, there are gaps in his approaches and the works of J. Visman and C. O. Ingamells definitely provide a powerful way to fill these gaps.

Pierre Gy breaks the total uncertainty manifestations into eight sources of sampling variability, which he called sampling errors; they are the Fundamental Sampling Error (FSE), the Grouping and Segregation Error (GSE), the Increment Delimitation Error (IDE), the Increment Extraction Error (IEE), the Increment Weighting Error (IWE), the Increment Preparation Error (IPE), plus a long-range error and a periodic error for dynamic measurements. Other sources of variability should be added such a laboratory analytical error and the in-situ nugget effect (e.g., for geologists and geochemist), which is indeed a problem for sampling nonbroken materials.

MU experts and empiricists seem to resent such classification as they think the conventional statistical analysis of data alone is enough to detect sampling problems; in this, there is a firm belief that all variability can be modelled by two systematic components (bias, acceptable or not) and a stochastic variance (precision, acceptable or not). However, and this is critically important, detection is not identification, and most certainly not cure; obviously something of great importance is missing. As a result, MU should welcome TOS because it effectively pinpoints where problems are, and this is the cardinal issue that gives irrefutable solutions to minimize each source of excessive sampling variability.

As a quick reminder, TOS makes a clear distinction between Uncertainty (i.e., after all sources of sampling biases have been minimized to a negligible level, and after precision has been reduced to an acceptable level relative to preselected logical objectives), and Error when no attempts to minimize sampling correctness problems and unacceptable precision are made. The word Error was selected by Pierre Gy because at the time, early sixties, in an overwhelming amount of cases, sampling incorrectness and excessive precision problems were the rule rather than the exception, which has nothing to do with uncertainty and everything to do with major mistakes. These definitions in TOS may bother some MU experts, but I do not think they are totally incompatible with their ways of thinking. Therefore, it would be beneficial through positive thinking to negotiate intelligently, within WCSB if possible. This third edition of my book is my contribution into such a critically important endeavor.

After 40 years of involvement in the Theory of Sampling I owe my knowledge to many experts, many friends, some of them long gone: for those who are gone, they did not die, but they simply faded away in a superb intellectual glow.

First, I thank these extraordinary beings who I wish would still be here today, as they helped me so much. Charles Oliver Ingamells taught me the wonderful world of Poisson processes, the classic chemical methods of analysis, and the secrets of preparing reliable Reference Materials. He broke many paradigms I floundered with, setting me free for who I am today. Dr. Michel David, a famous geostatistician, with whom I shared many ideas on sampling, taught me the fundamentals in Geostatistics at Colorado School of Mines (CSM), and encouraged me to teach sampling courses at the continuing education department of CSM: this was the very beginning of my success in 1987. Ever since, every year in October, I have taught a sampling course at CSM for the industry; the course acquired a worldwide reputation.

My thanks to Dr. Pierre M. Gy are limitless. He was my true mentor, a good man and a good friend. From the very beginning that I was introduced to his work not only did I develop a passion for his massive creativity, but I also admired his ethics as a good, caring human being. It has always been my opinion that Pierre deserved far more from the academic world than what he was given over the years, and, above all, should have been given the highest honors of the land from his government. What kind of man can be a master in mineral processing, mathematics, statistics, and create the Theory of Sampling? His work must never die, and I take it as my personal mission that it never will die.

Dr. Dominique François-Bongarçon, president of Agoratek International, was for many years my sampling-expert competitor, and we often crossed each other’s path in our consulting and teaching practices around the world. I consider Dominique as a bright expert showing a well-balanced knowledge of the TOS and of Geostatistics. Dominique’s work is the best available today to offer a clear, essential link between the TOS and Geostatistics, and I would like to extend that link to Chronostatistics. I am t...

Table of contents

• Cover
• Half Title
• Title Page
• Copyright Page
• Table of Contents
• Foreword
• Preface
• Author’s biography
• Part I: Introduction and a Management Strategy
• Part II: Fundamental Statistical Concepts Used in the Theory of Sampling
• Part III: Heterogeneity and Homogeneity
• Part IV: Sampling Errors Introduced by Various Forms of Heterogeneity
• Part V: Integration of Visman and Ingamells’s Works into the Theory of Sampling
• Part VI: The In-Situ Nugget Effect A Major Component of the Random Term of a Variogram
• Part VII: The Capital Notion of Sampling Correctness
• Part VIII: The Increment Weighting Error and the Weighing Error
• Part IX: Review of Some Notorious Sampling Problems
• Part X: Chronostatistics
• Part XI: Homogenization
• Part XII: Recommendations to Manufacturers of Sampling Equipment and to Engineering Firms
• References
• Index