<jats:title>Significance</jats:title>
<jats:p>
Kimberlites are igneous rocks derived from deep mantle sources. Recent studies have suggested that certain kimberlites originated from a mantle source with relatively primitive chemical composition that was created by early Earth processes. We present W isotope data for a global suite of kimberlites with variable formation ages and find their mantle source(s) to be characterized by
<jats:sup>182</jats:sup>
W/
<jats:sup>184</jats:sup>
W averaging ∼6 ppm lower than the upper mantle ratio. This result is consistent with derivation of some kimberlites from one or more early formed mantle reservoirs. The low
<jats:sup>182</jats:sup>
W/
<jats:sup>184</jats:sup>
W of these kimberlites is indicative of an ancient mantle source modified by some form of core–mantle interaction, an early silicate fractionation event, an overabundance of late-accreted materials, or a combination of these.
</jats:p>

<jats:p>In recent years, the <jats:sup>187</jats:sup>Re–<jats:sup>187</jats:sup>Os isotope system has been increasingly used to study samples containing very small quantities of Os. For such samples, optimisation of measurement procedures is essential to minimise the loss of Os before mass spectrometric measurements. Micro‐distillation is a necessary purification step that is applied after the main Os chemical separation procedure, prior to Os isotope ratio measurements by negative‐thermal ionisation mass spectrometry (N‐<jats:styled-content style="fixed-case">TIMS</jats:styled-content>). However, unlike the other separation steps, this procedure has not yet been optimised for small samples. In this study, we present a refined micro‐distillation method that achieved higher yields and allowed high‐precision R(<jats:sup>187</jats:sup>Os/<jats:sup>188</jats:sup>Os) expressed as <jats:sup>187</jats:sup>Os/<jats:sup>188</jats:sup>Os measurements for small‐sized geological samples that contain only a few pg Os. The Os recovery in the micro‐distillation step was tested by changing the operating conditions including heating time and temperature, and amounts of oxidant and reductant. Recoveries were measured by the isotope dilution <jats:styled-content style="fixed-case">ICP</jats:styled-content>‐<jats:styled-content style="fixed-case">MS</jats:styled-content> method after the addition of <jats:sup>190</jats:sup>Os‐enriched spike solution. We found that the most critical factor controlling the chemical yield of Os during micro‐distillation is the extent of dilution of the reductant (<jats:styled-content style="fixed-case">HB</jats:styled-content>r) by H<jats:sub>2</jats:sub>O evaporated from the oxidant. A refined micro‐distillation method, in which the amount of oxidant solution is reduced from the conventional method, achieved an improved chemical yield of Os (~ 90%). This refined method was applied to the measurement of <jats:sup>187</jats:sup>Os/<jats:sup>188</jats:sup>Os by N‐<jats:styled-content style="fixed-case">TIMS</jats:styled-content> of varying test portions of the geological reference material <jats:styled-content style="fixed-case">BIR</jats:styled-content>‐1a. The resulting <jats:sup>187</jats:sup>Os/<jats:sup>188</jats:sup>Os ratios of <jats:styled-content style="fixed-case">BIR</jats:styled-content>‐1a matched the literature data, with propagated uncertainties of 0.2, 1.1 and 11% digested sample quantities containing 150, 10 and 1 pg of Os, respectively.</jats:p>