Facere, the Latin root of 'fact', means to make or to do; hence, factory, manufacture, and artifact.
Here's the open-access edition of the OED on the origin of the English word 'fact':
Late 15th century: from Latin factum, neuter past participle of facere 'do'. The original sense was 'an act', later 'a crime', surviving in the phrase before (or after) the fact. The earliest of the current senses ( 'truth, reality') dates from the late 16th century.[At the above link, the OED defines 'fact', as that term is used today, in an epistemological fashion, defining it as 'a thing that is known (or proved) to be true' (Oxford Dictionaries, Oxford University Press. http://www.oxforddictionaries.com/definition/english/fact )].
In Samuel Johnson's 1755 dictionary, 'fact' is still defined as an act, but now in opposition to supposition and speculation. Johnson's first definition is, 'A thing done; an effect produced; something not barely supposed or suspected, but really done.' His second definition of 'fact' reads as follows: 'Reality; not supposition; not speculation' (Johnson "Fact." A Dictionary of the English Language: A Digital Edition of the 1755 Classic by Samuel Johnson, ed. Brandi Besalke). These two definitions don't explicitly give an epistemological rendering of 'fact'; the 'thing done' or 'reality' can remain unknown. Still, in view of some of Johnson's samples of the word's usage, it looks like he had in mind an epistemological interpretation for at least some uses of 'fact'; for him, 'fact' at least sometimes means known or (at least) evident fact.
Notice that a fact was a particular action or 'thing done'. This is crucial, for it conditions the sorts of evidence that are to be sought when determining the facts. To use Whately's example (from an earlier post), if we want to know whether someone is in the process of killing Alexander, we should look and see whether this is happening. Direct observation is the best way to settle the matter. If the killing is alleged to have already happened, then we should seek witnesses and request their testimony. Did they observe the killing? Of course, it's better to have several witnesses independently corroborating each others' testimony. If we can't reach the witnesses, we'll look for any documentation (signed statements, etc.) that they might have left behind concerning Alexander's demise.
Of course, these kinds of support are very like the types of evidence that are considered in jury trials, in which jurors are called upon to make determinations concerning matters of fact. Barbara J. Shapiro describes the evolution of the rules of evidence for such trials in late 15th-Century and early 16th-Century England.(Shapiro, Probability and Certainty in Seventeenth-Century England: A Study of the Relationships Between Natural Science, Religion, History, Law, and Literature [Princeton University Press, 1983], Ch. 5) . Shapiro argues that after the advent of these rules, the phrase 'matter of fact' migrated from law, where it pertained to observable human actions, to the natural sciences, where it took within its ambit the observable natural phenomena. Shapiro writes that the phrase 'matter of fact'
had traditionally been associated with the human facts of the historian or the law court. During the seventeenth century, 'matter of fact', like 'experience', would be assimilated into natural science. (Shapiro, Probability and Certainty in Seventeenth-Century England, p. 20)As Shapiro says, prior to the 17th Century matters of fact were central not only in legal inquiries but in historical studies, too. This old use of the phrase is evident in the following remark by Francis Bacon:
This facility of credit and accepting or admitting things weakly authorised or warranted, is of two kinds according to the subject: for it is either a belief of history (as the lawyers speak, matter of fact); or else of matter of art and opinion. (Bacon, 'The Advancement of Learning', Book I, iv, paragraph 9, 1605)Bacon here expands the range of history to take in not only human but, also, natural history. According to Shapiro,
Bacon's combination of the historico-legal 'fact' of human action with the natural fact established by observation and experiment made it possible for his successors to apply a familiar legal technique of verifying events in the human world to natural phenomena. (Shapiro, Culture of Fact: England, 1550–1720 [Cornell University Press, 2000], p. 109)While other historians of science (e.g., Lorraine Daston and Steven Shapin) may not attach as much influence to Bacon as Shapiro does, I take it that they generally agree about the meaning of 'matter of fact' in the 17th Century. At that time, matters of fact were particular, observable occurrences, for which the best evidence was observation. If one could not, oneself, observe the event in question, then one ought to rely on the testimony of those who had witnessed the event. One should seek corroboration by multiple witnesses. Matters of fact did not include the physical laws of nature. They did not include well-grounded theories about the underlying causes of the observed phenomena. They did not include dispositions or causal powers. Nor did they include any religious truths or truths about moral and aesthetic values that there might be. It's in that light that we should read this bit from John Donne:
Contemplative and bookish men must of necessity be more quarrelsome than others, because they contend not about matter of fact, nor can determine their controversies by any certain witnesses, nor judges. But as long as they go towards peace, that is Truth, it is no matter which way. (Donne, Biothanatos,'Preface', 1608)In denying that the disputes among bookish characters were disagreements over matters of fact, Donne did not mean that the the disputes were mere emotional venting contests with no truths at stake. There were truths at stake, so that some of the disputants were right and some were wrong. Some held true beliefs and others had false ones. Donne meant only that the kinds of truths at stake were not of the relatively straightforward sort that could be settled just by looking (or by relying on witness' testimony).
The 17th-Century use of 'matter of fact' is also exhibited in this passage from Locke's Essay:
Probability is either of sensible matter of fact, capable of human testimony, or of what is beyond the evidence of our senses. But to return to the grounds of assent, and the several degrees of it, we are to take notice, that the propositions we receive upon inducements of probability are of two sorts: either concerning some particular existence, or, as it is usually termed, matter of fact, which, falling under observation, is capable of human testimony; or else concerning things, which, being beyond the discovery of our senses, are not capable of any such testimony. (Locke, Essay Concerning Human Understanding, Book IV, ch. xvi, paragraph 5, 1689 [Emphases added])The 17th-Century sense of 'matter of fact' survived at least till the middle of the 19th Century, where we find its exclusion of many scientific truths being emphasized by Sir George Cornewall Lewis in his 'Essay on the Influence of Authority in Matters of Opinion' (1849):
By a matter of fact, I understand anything of which we obtain a conviction from our internal consciousness or any individual event or phenomenon which is the object of sensation. ... A fact, as so defined, must be limited to individual sensible objects, and not extended to general expressions or formulas, descriptive of classes of facts, or sequences of phenomena, such as that the blood circulates, the sun attracts the planets, and the like. Propositions of this sort, though descriptive of realities, and therefore, in one sense, of matters of fact, relate to large classes of phenomena, which cannot be grasped by a single sensation, which can only be determined by a long series of observations, and are established by a process of intricate reasoning. (Lewis, 'Essay on the Influence of Authority in Matters of Opinion', 1849, pp. 1-2)Here, matters of fact end where reasoning begins, and facts are the data or raw material on which reason operates. In short, matters of fact were just those questions that could be settled by eyewitness testimony, without requiring any expert witnesses (that is, without requiring the witness to be especially adept at physics or medicine, etc.). A similar conception of matters of fact is evident in the 7th, revised edition of Bishop Whately's Elements of Rhetoric (1846), where Whately wrote:
If, for instance, a person ... states, that in the East Indies he saw a number of persons who had been sleeping exposed to the moon's rays, afflicted with certain symptoms, and that after taking a certain medicine they recovered, he is bearing testimony to simple matters of fact: but if he declares ... that the patients in question were so affected in consequence of the moon's rays, that such is the general effect of them in that climate, and that that medicine is a cure for such symptoms, it is evident that his testimony ... is borne to a different kind of conclusion; namely, not an individual, but a general conclusion, and one which must rest, not solely on the veracity, but also on the judgment, of the witness.(Whately, Elements of Rhetoric, 7th edition, ch. II, sec. 4, 1846, pp. 53-4 [p. 61 in the English Fellowes edition] [Emphases in the original])Clearly, Lewis and Whately did not mean to limit the domains of inquiry in which we may legitimately seek a truth of the matter to just the domain of matters of fact. Like Donne, they would take there to be a truth of the matter in disputes about causal claims and in disputes over moral values. They would regard some people as holding false beliefs about causal relations and moral values. They would also allow for the use of reason in trying settle such disputes or in trying to arrive at the truth in such inquiries. All of this is compatible with their denying that these areas of inquiry concern matters of fact; for in their day, 'matters of fact' pertained only to those questions that could be settled by observation (or by the testimony of someone who had made the relevant observations).
I think that, nowadays, we're more likely to take 'matters of fact' as being co-extensive with topics for which there is a truth of the matter, topics about which we may hold true or false beliefs (and about which we can make progress by reasoned inquiry). If we combine this more recent interpretation of 'matters of fact' with the old idea that matters of fact are just those that can be settled by a single observation, then we'll arrive at a naive kind of positivism, one which regards disagreements about value as so much hot air, or emotional venting, mere seeming disputes for which there is no truth of the matter. Perhaps such a move led to some of the flawed, modern versions of the fact-opinion distinction that are being taught in some public schools.