©2016 Society of Economic Geologists, Inc. Economic Geology, v. 111, pp. 503–525 Using Microdiamonds in Kimberlite Diamond Grade Prediction: A Case Study of the Variability in Diamond Population Characteristics Across the Size Range 0.2 to 3.4 mm in Misery Kimberlite, Ekati Mine, NWT, Canada* M. Y. Krebs, 1,† D. G. Pearson, 1 T. Stachel, 1,2 R. A. Stern, 1,2 T. Nowicki, 3 and S. Cairns 4 1 Department of Earth and Atmospheric Sciences, University of Alberta, Edmonton, Canada 2 Canadian Centre for Isotopic Microanalysis, University of Alberta, Edmonton, Canada 3 Mineral Services Canada, Vancouver, BC, Canada 4 Northwest Territories Geoscience Ofice, Yellowknife, NT, Canada Abstract First predictions of the macrodiamond grade of newly discovered kimberlites are commonly obtained using size frequency distributions of microdiamonds. The success of this approach suggests a common origin of microdiamonds and macrodiamonds, an implication not yet conclusively established or disproved. In contrast to previous comparative studies on microdiamonds and macrodiamonds from single deposits, here all diamonds analyzed originate from the same microdiamond samples (558 diamonds, ranging from 0.212 to 3.35 mm). The diamonds were analyzed for their carbon isotope compositions and nitrogen characteristics, and, based on this dataset, statistical comparisons were conducted across the size range to assess cogenesis. As a whole, the Misery diamond suite shows high nitrogen contents (median = 850 at. ppm), a bimodal distribution in time-averaged mantle residence temperatures (two distinct subpopulations in mantle residence temperatures: ≤1,125° and ≥1,175°C), a high degree of platelet degradation, and δ 13 C compositions that are isotopically slightly heavier (median = –4.4‰) than the global median. Statistical comparisons of the various size classes indicate the presence of subtly different subpopulations at Misery; however, the nature and magnitude of these geochemical differences are very small in the context of the global diamond database and are viewed as petrogenetically insigniicant. The general geochemical similarity of diamonds from different size fractions at Misery reinforces the use of size-frequency analysis to predict diamond grade in kimberlite diamond deposits. Introduction In diamond exploration, the evaluation of newly discovered kimberlites and lamproites consists of several stages (e.g., Rombouts, 1995, 1999). The irst direct assessment of dia- mond content is usually recovery of microdiamonds from rel- atively small samples (~20–250 kg) in the form of drill core, grab samples, or material from pitting and trenching. These rock samples then undergo caustic fusion and, subsequently, all diamonds are recovered by visual sorting of the residues. First used by Deakin and Boxer (1989) on diamonds from the Argyle mine (Western Australia), an initial prediction of the macrodiamond grade (carats/tonne) can then be obtained using size frequency distributions. These size distributions can be expressed as a weight or number frequency distribu- tion versus diamond diameter (sieve size) or weight (in car- ats). For microdiamond samples, sieve size is usually chosen, as diamonds in a typical exploration sample are too small and numerous to make weighing of individual diamonds practical (Rombouts, 1995). On a log-log plot, the frequency of micro- diamonds and the sieve class display a near-linear correlation. A shallow gradient combined with a high number of larger (heavier) diamonds indicates a higher economic potential than a curve with a steep gradient and/or low diamond count per unit size/weight (Boxer, 2004). There are two published methods of extrapolating these size distributions to estimate the grade of macrodiamonds; one is based on the log-normal model (Rombouts, 1997), the other on a Pareto tail model (see Rombouts, 1999, for details). While the size-frequency distributions and the coeficients deining their linear rela- tionship for microdiamonds can be used to make an early assessment of the expected diamond content, it is believed that quantitative grade estimates of macrodiamonds on the basis of microdiamonds only are not achievable (Chapman and Boxer, 2004). If the results of a microdiamond analysis are favorable, i.e., indicating a recoverable macrodiamond (typi- cally >1.8 mm) grade above 0.1 carats/tonne, the target will move on to the next stage of assessment. Although microdiamond evaluation has led to misinter- pretation of deposit grade in some cases (e.g., Pattison and Levinson, 1995), the approach generally works very well as an exploration tool and is currently in widespread use. The resulting size distributions are based solely on statistics and the success of the method implies a common origin of micro- and macrodiamonds. This implication of cogenesis has been addressed in several comparative studies investigating dif- ferences in diamond morphology, mineral inclusion chemis- try, N characteristics, and δ 13 C compositions of micro- and macrodiamonds; however, is has not yet been conclusively established or disproved (e.g., Haggerty, 1986; McCand- less et al., 1994; Pattison and Levinson, 1995; Kaminsky and Khachatryan, 2001; Sobolev et al., 2004; Johnson et al., 2012; Melton et al., 2013). McCandless et al. (1994), for example, suggested a common origin, whereas several other studies found evidence that microdiamonds may represent separate diamond populations produced partially or entirely by differ- ent diamond-forming processes or in distinct mantle sources 0361-0128/16/4387/503-23 503 Submitted: February 10, 2015 Accepted: October 8, 2015 † Corresponding author: e-mail, krebs@ualberta.ca *A digital supplement to this paper is available at http://economicgeology.org/ and at http://econgeol.geoscienceworld.org/.