ABSTRACT
Many experiments are very expensive to conduct so that ex-perimenters are often forced to limit the number of treatment combinations that can be studied in order to have adequate resources available to replicate the treatment combinations under study. The replication of treatment combinations is necessary in order to be able to have an independent estimate of the experimental error variance, which is denoted by o2 in this book. Having a good estimate of o2 is very important when the major objective of the experiment is confirmatory. However, if the major objective of the experiment is exploratory, it is often more desirable to study many different treatment combinations, each performed once, rather than a few treatment combinations each replicated many times. This book is devoted to methods that can help to extract the relevant information in experiments that are not replicated.
