Salmon’s article describes the first of several alternatives to the deductive-nomological picture of explanation that we will be looking at. He also introduces some of the issues that arise concerning explanation in any case where we have no more that statistical or probabilistic information to go on.
• Salmon begins with a discussion of Hempel’s approach to this sort of case, an approach by way of his idea of “inductive-statistical” (I-S) explanation (explanation using statistical laws in an argument that is inductive in the sense that its premises may render its conclusion no more than highly probable). Salmon’s discussion in this part of the paper involves some terms that may need more explanation than he gives, at least when he first uses them.
• The “requirement of total evidence” mentioned on p. 209 is a response to the fact (noted by Russell in his paragraph 6.17) that adding information can reduce the probability of an inductively supported conclusion. The requirement is that, to count an inductive argument as strong, you must consider as premises the totality of the information you have available.
• The related “requirement of maximal specificity” mentioned on p. 210 is explained in the first paragraph of p. 211.
• “Statistically relevant” information is information that effects the probability of something. Hempel gives an example explaining the idea in the last full paragraph of p. 211.
• Most of the second half of the paper is devoted to a series of examples illustrating Salmon’s approach to explanation. You should think not only whether Salmon gives the correct account of statistical explanation of single events but also whether it is even correct to suppose there are explanations in some of these cases. What he refers to as “Principle 1” (see p. 213) is relevant here, and you should think about the way Salmon frames his objection to this principle in his last paragraph.