Abstract This paper describes the petrology, geochemistry and petrogenesis of Archean granitoids. Archean granites define a continuum of compositions between several end members: (i) magmas that originated by partial melting of a range of crustal sources, from amphibolites to metasediments (‘C-type’ granitoids); and (ii) magmas that formed by partial melting of an enriched mantle source, the most common agent of enrichment being felsic (TTG) melts. Differences in the degree of metasomatism results in different primitive liquids for these ‘M-type’ granitoids. Mixed sources, differentiation and interactions between different melts resulted in a continuous range of compositions, defined by variable proportions of each end member. During the Archean, evolved crustal sources (sediments or felsic crust) and metasomatized mantle sources become increasingly more important, mirroring the progressive maturation of crustal segments and the stabilization of the global tectonic system.

This chapter considers the tests and procedures used to describe or evaluate the physical, mechanical and chemical characteristics of aggregates, for the purposes of (a) prediction of the likely ‘in service’ behaviour of the material (b) comparison between competing materials (c) specification compliance or (d) quality control. Individual limits are not discussed here but are considered in the appropriate Chapter. Sub-base materials (Chapter 10 ) sometimes have to use tests in BS 1377, but this section confines itself to BS 812 and the relevant standards from other countries. The first step is the collection of samples. Statistically, a sample can be defined as an individual or group of individuals drawn from a large or infinite population, Information obtained from samples is only as representative of the material as the samples on which they are performed . If observations reveal little variation and there has been no bias in collecting, then a small sample or small number of samples may be highly representative of a population. If the variation is large then more and/or larger samples will be required Representative sampling, however, is perhaps the most difficult of the control operations to perform satisfactorily. Sampling, as with all types of test, introduces sources of variation and error, so that judgements of materials based on infrequent random tests are fraught with difficulties. In this connection, see also the remarks on sampling in relation to classification given in Chapter 6 .A random sample is one in which each potential observation has an