Mervyn Stone died on 19 September 2020, aged 87. He was a brilliant mathematician, professor of probability and statistics, and thinker.

He was elected to the Royal Statistical Society in 1955; served on the Series B Editorial Panel (1966–1969) and as the Editor of Series B (1975–1977); on the Research Section Committee (1974–1977), the Conference Committee (1977–1978) and as a Member of Council (1976–1980). He was awarded the Guy Medal in Silver in 1980 for his contributions to statistical theory. His theoretical interests are listed in his curriculum vitae as criticism of formal Bayesian methods, design of experiments, large deviations, cross‐validation and coordinate‐free multivariate analysis. His applied work included applications in psychology, pharmacology, stem cell modelling, water privatisation and the influence of darkness on road casualties. He was passionate about statistics and its use to improve society. After he retired, he undertook a number of projects related to the use, and misuse, of statistics in public policy, including funding of the National Health Service (NHS), performance of the police service, traffic safety measures and immigration. He made lasting contributions to statistics and to our society.

The ‘eventually renowned teachers’ included Henry Daniels and Dennis Lindley. In that year, he also met Solveig, his wife‐to‐be.

After his Diploma, Mervyn accepted a job at the MRC Applied Psychology Research Unit in Cambridge where he obtained permission to work part time for a PhD, in which he chose to study the application of Shannon’s information measure to the design and comparison of regression experiments. He completed his PhD in Statistics in 1958 and then took up a 1‐year post as a Fulbright research associate at Princeton before being appointed to a lectureship in 1961 in Dennis Lindley’s new statistics department at the University College of Wales at Aberystwyth. He spent the year 1965–1966 as a visiting professor at the University of Wisconsin, then took up a Readership at the University of Durham, before moving to University College London in 1968, first as a Reader then as Professor of Probability and Statistics, and, later, Head of Department. Dennis Lindley had recently been appointed to the Chair in Statistics and had a vision of promoting the Bayesian philosophy. Dennis was delighted to appoint Mervyn, who, he told us, was a Bayesian. But Mervyn was far too free‐thinking to be limited by that epithet and in fact very little of Mervyn’s work in statistical theory was directed towards developing formal Bayesian methods.

There was a special Biometric Society meeting in London in the early 1970s where David Cox and Dennis Lindley were called upon to argue the cases, respectively, for and against the use of randomisation in statistical inference and design. Mervyn proposed the vote of thanks to the speakers and in an eloquent contribution said that he wholeheartedly agreed with Dennis’s argument but with David’s conclusion.

Mervyn’s research papers cover a wide range of mathematical methods and theoretical ideas and are notable for their originality at many levels, as well as a ‘minimalist’ style of writing. His publications also include a variety of scientific applications, authoritative studies on Florence Nightingale and Adolphe Quetelet, papers on the use of statistics in public finance and his book Coordinate‐free Multivariable Analysis, subtitled ‘An illustrated geometric progression from Halmos to Gauss and Bayes’ (Clarendon, 1986). In addition to his notable discussion paper, ‘Strong inconsistency from uniform priors’ (in the Journal of the American Statistical Association, 1976), which includes his famous Flatland example introducing what is now known as Stone’s paradox, he frequently presented papers for discussion at Royal Statistical Society meetings. Among these, ‘Marginalization paradoxes in Bayesian and structural inference’ (with Philip Dawid and James Zidek, in the Journal of the Royal Statistical Society, Series B, 1973) was highly influential (causing Dennis Lindley to retract his own ideas on improper priors) and ‘Cross‐validatory choice and assessment of statistical predictions’ (in the Journal of the Royal Statistical Society, Series B, 1974) spawned what is now a far‐reaching methodology.

Mervyn was a dedicated teacher who strove to find new ways to impart understanding. As one example, in his ‘Pebble sampling experiment’ students were asked to estimate the total weight of 100 pebbles by sampling and weighing 10 of them, using a variety of sampling schemes. They were stones of different shapes and sizes that he had collected from a river bed in Wales. This simple exercise turned out to be remarkably effective in teaching concepts of probability, estimation, sampling distributions, bias and variance, as well as the sampling schemes themselves, and it is still used today.

Aspects of these projects were brought together in a thoughtful booklet Failing to Figure: Whitehall’s Costly Neglect of Statistical Reasoning (Civitas, 2009), which concluded with six suggestions for improving policy‐making.

He was devastated by the sudden death in 1994 of his son Richard, aged 32 with a young family and a promising career in statistics, caused by a rare genetic condition, and again in 2008 by the untimely death from cancer of his wife Solveig. He is survived by his daughter Helen and five grandchildren.

默文·斯通（Mervyn Stone）于2020年9月19日去世，享年87岁。他是一位杰出的数学家，概率论和统计学教授以及思想家。

他当选为英国皇家统计学会于1955年; 曾任职于B系列编辑小组（1966–1969）和B系列编辑（1975–1977）；研究委员会（1974–1977），会议委员会（1977–1978）和理事会成员（1976–1980）。由于对统计理论的贡献，他在1980年获得了银盖伊奖章。他的理论兴趣列在他的简历中作为对正式贝叶斯方法，实验设计，大偏差，交叉验证和无坐标多元分析的批评。他的应用工作包括心理学，药理学，干细胞建模，水私有化以及黑暗对道路伤亡的影响。他对统计数据及其对改善社会的使用充满热情。退休后，他从事了许多与公共政策中的统计使用和滥用有关的项目，包括国家卫生服务局（NHS）的资金，警察部门的绩效，交通安全措施和移民。他为统计和我们的社会做出了持久的贡献。

“最终著名的老师”包括亨利·丹尼尔斯和丹尼斯·林德利。那年，他还遇到了即将成为妻子的索尔维格。

在获得文凭后，默文恩在剑桥的MRC应用心理学研究部门接受了一份工作，在那里他获得了兼职博士学位的许可，其中他选择研究Shannon信息量度在设计和比较回归实验中的应用。他于1958年完成了统计学博士学位，然后在普林斯顿大学担任富布赖特（Fulbright）研究助理一年，之后于1961年被任命为威尔士大学阿伯里斯特威斯大学丹尼斯·林德利新统计学系的讲师。1965年至1966年，他在威斯康星大学任客座教授，之后在达勒姆大学获得了读者身份，之后于1968年进入伦敦大学学院，先是担任读者，然后是概率与统计教授，以及，后来，系主任。丹尼斯·林德利（Dennis Lindley）最近被任命为统计学主席，并具有推广贝叶斯哲学的远见。丹尼斯很高兴任命默文（Mervyn），他告诉我们，他是贝叶斯主义者。但是，默文（Morvyn）太自由了，以为不受该称谓的限制，事实上，默文在统计理论方面的工作很少涉及开发正式的贝叶斯方法。

1970年代初在伦敦召开了一次特别的生物识别学会会议，要求戴维·考克斯（David Cox）和丹尼斯·林德利（Dennis Lindley）分别就支持和反对在统计推断和设计中使用随机数论证。默文提议对发言人表示感谢，并雄辩地表示，他完全同意丹尼斯的论点，但同意大卫的结论。

默文的研究论文涵盖了广泛的数学方法和理论思想，并且在许多层面上都独具匠心，并且具有“极简主义”的写作风格。他的出版物还包括各种科学应用，对佛罗伦萨·南丁格尔和阿道夫·奎特莱特的权威研究，关于在公共财政中使用统计数据的论文以及他的著作《无坐标多变量分析》，标题为“从Halmos到高斯和贝叶斯的图解几何级数”。 （Clarendon，1986）。除了他著名的讨论论文“统一先验的强烈矛盾”（在《美国统计协会杂志》上），1976年），其中包括他著名的Flatland例子，介绍了现在所说的斯通悖论，他经常在皇家统计学会会议上发表论文供讨论。其中，“贝叶斯边际化悖论和结构推断”（与Philip Dawid和James Zidek于1973年在皇家统计协会期刊B系列上合影）具有很大的影响力（使Dennis Lindley收回了关于不当先验的自己的观点）和“交叉验证性选择和统计预测的评估”（《皇家统计学会杂志》，B系列，1974年）催生了如今意义深远的方法论。

默文（Mervyn）是位敬业的老师，他竭尽所能寻找传授理解的新方法。举一个例子，在他的“鹅卵石采样实验”中，要求学生通过使用各种采样方案对10个鹅卵石进行采样和称重，以估算100个鹅卵石的总重量。这些石头是他从威尔士河床采集的不同形状和大小的石头。事实证明，这种简单的练习在教授概率，估计，抽样分布，偏差和方差以及抽样方案本身的概念方面非常有效，并且至今仍在使用。

这些项目的各个方面都汇集在一本深思熟虑的小册子中：《未能图：白厅对统计推理的昂贵忽视》（Civitas，2009年），最后提出了六项改善政策制定的建议。

1994年，他的儿子理查德（Richard）突然去世，他因32岁的罕见家庭遗传病而突然去世，他因年轻的家庭和令人鼓舞的统计职业而丧命。2008年，他的妻子索尔维格（Solveig）因癌症过早去世，再次使他感到震惊。他的女儿海伦（Helen）和五个孙子幸存下来。