There are some things that can only be known by acquaintance, i.e. by “direct experience”. If you want to know what red is, it’s no use having it described to you; you have to experience it for yourself. Such knowledge is not propositional: it is knowledge concerning a thing (what kind of thing is a point of contention). Typically, we know what certain qualities are in this way—and this is not a matter of knowing that such-and-such. The question I am concerned with now is the transmissibility of such knowledge—the possibility of conveying it to someone hitherto ignorant of it. For there is a marked contrast between knowledge by acquaintance and knowledge by description in this regard: the latter can be conveyed to someone not already in possession of it, while the former cannot. If I have acquaintance knowledge of red, I cannot communicate this knowledge to you by the usual methods: by means of language or by acting in certain ways. But if I have a piece of propositional knowledge (knowledge of a fact) I can transmit it to you. Even if the knowledge concerns an inner state of mine, as private as you wish, I can let you know what the fact is that I know: I can cause you to know the same fact–for example, the fact that I am in pain. This knowledge is transmissible, but I can’t transmit my knowledge of what pain is—I can’t cause you to have it by verbal or behavioral means. I can’t cure your ignorance by communicating my knowledge to you: I can’t teach you what pain is; this you have to know for yourself. Pain teaches you what pain is; other people can’t. But in the case of propositional knowledge it is always teachable: it can be transferred from person to person. We may thus venture the following generalization: all propositional knowledge is transmissible, but no acquaintance knowledge is transmissible. We could even strengthen this generalization as follows: all propositional knowledge must be transmissible, but no acquaintance knowledge could be transmissible. You can be educated about any fact, but you can never be educated about whatever it is that acquaintance knowledge is knowledge of. No one can instruct you about what red is; only experience can—by vouchsafing you a sensation of red.
This distinction should be distinguished from other distinctions in the rough neighborhood. It is clearly quite distinct from the a priori-a posteriori distinction: acquaintance knowledge is typically (though not always) a posteriori, and some propositional knowledge is a priori. Nor does it coincide with the innate-acquired distinction: acquaintance knowledge could be either, as could propositional knowledge. Nor is it the same as the distinction between basic and derivative knowledge. Nor is it the same as the distinction between subjective and objective knowledge. It would also be wrong to say that it coincides with the public-private distinction: acquaintance knowledge can be shared by many people (nearly all of humanity), and not all propositional knowledge need be publicly possessed. It’s also not the same as the certain-uncertain distinction. The distinction I am making is specifically about transmissibility: what can be conveyed, taught, passed on to others. Propositional knowledge can always in principle be transmitted from one person to another, so that it doesn’t need to be known first-hand; but acquaintance knowledge can only be acquired from one’s own resources—it can’t be outsourced. No lessons, no matter how persistent or expensive, will ever inculcate knowledge of what red is; it must be acquired by the individual acting alone. Of course, you could sign up for a treatment that caused experiences of red in you if you naturally lack them: this would be someone else causing you to have knowledge of what red is. But it isn’t the same as someone verbally explaining to you what red is, as they might explain to you what the equator or iambic pentameter is. The general point is that knowledge comes in two types, the transmissible and the non-transmissible. We could say that the former type can always be possessed by testimony while the latter type can never be. This is by no means a trivial observation and raises the difficult question of why it should be so.
What about knowing-how—is it transmissible or not? Evidently it is: a skill can be conveyed from one person to another by verbal instruction or by example. Of course the student must have capacities sufficient to learn from the instruction, but that is also true of transmitting propositional knowledge. And some will be better at acquiring the knowledge than others—also true of propositional knowledge. Still, the knowledge can be inculcated from afar (aided by individual practice). So acquaintance knowledge stands out from the other main kinds of knowledge in respect of its non-transmissibility: it is the odd one out. Except notice this fact: propositional knowledge presupposes acquaintance knowledge. Recall Russell’s famous pronouncement: “Every proposition which we can understand must be composed wholly of constituents with which we are acquainted”. We need not accept the “wholly” here, but the point is not lost: transmissible knowledge is built up from non-transmissible knowledge. But not vice versa: so, as Russell says, acquaintance knowledge is more basic than propositional knowledge. There is no propositional knowledge of facts without acquaintance knowledge of things. We therefore could not transmit knowledge unless we had knowledge we can’t transmit. Even the most transmissible of knowledge rests upon knowledge that resists transmission (save by causing suitable experiences in the recipient). Thus non-transmissible knowledge is as valuable and essential as the transmissible kind—the kind we prize in science. Even physics presupposes knowledge that cannot be conveyed from one person to another (mathematics too if we have acquaintance with mathematical entities). If there were no such knowledge, transmissible scientific knowledge would be impossible. When Frege talked about science passing knowledge on down through the generations via Thoughts he was forgetting that this is only possible because some knowledge can’t be so passed on: it must be discovered afresh by every generation. You can only learn from your teachers because there are things you can’t learn from them—by those mysterious self-generated acts of acquaintance. If teachers had to instill knowledge by acquaintance in you, you would never learn anything—and they would be out of a job. Education can only instill knowledge because it is not the only thing that instills knowledge. Educators might benefit from reflecting on this.
Epistemology changes once this point is fully absorbed. There is not simply a homogeneous body of transmissible knowledge—knowledge of propositions. There are two categories of knowledge that work quite differently, one of them capable of passing from mind to mind, the other resistant to such passage. One kind of knowledge can be imported from outside while the other has to be homemade. The kind that can be transmitted consists of facts—objects having properties—but the kind that has to be homegrown concerns the nature of properties or types (universals, as Russell would say). We can also be acquainted with tokens (particulars) but our knowledge concerns the general type. Tastes provide a good example: you may know the taste of oysters but not be able to convey this knowledge to anyone else (except by offering them an oyster). This is quite different from knowing propositionally that there are six oysters on the table. Both types of knowledge coexist and commingle, as when I know that a certain restaurant specializes in oysters with a particular taste. The proposition can only be understood if knowledge of the taste of oysters is brought to bear on it. Epistemology thus needs two departments corresponding to the two types of knowledge. I think there has been a kind of prejudice against knowledge by acquaintance, possibly stemming from its essential incommunicability: real knowledge should not be so confined, it is thought. But actually it is the bedrock of all knowledge, just as Russell says: we rely on a natural convergence of acquaintance in our epistemic dealings, and if we didn’t have it communication would be impossible. There is a shared acquaintance with universals that everyone brings to the table, but this is not something that can be taught—as we know from the case of the blind. To put it differently, shared experience is the basis of all knowledge and all knowledge transmission (the grain of truth in empiricism). This is not shared “form of life”, as Wittgenstein would have it, but shared direct knowledge of universals (properties, types). In knowledge by acquaintance we are brought very close to the things known, and this is imported into all our knowledge, even concerning remote matters. This closeness is what cannot be duplicated by verbal instruction or observation of the other: experience of red tells me what red is in a way that nothing else can, and it is the basis of my grasp of propositions about the world of fact (“This apple is red”). What I have wanted to point out is that such knowledge, though vital, cannot be transmitted from person to person: if you don’t have it, tough luck. No amount of education can remedy your ignorance: if you don’t know what oysters taste like, you are going to have to eat one to find out, or remain forever ignorant on the subject.
 We can analyze propositional knowledge as true justified belief, but we can’t analyze acquaintance knowledge that way: it isn’t a type of belief at all (belief that what?), and the notion of justification has little purchase. It is really quite surprising that we use the word “know” for both cases, though entirely natural.
 The concept of knowledge by acquaintance is usually left at a rather intuitive level without much effort to investigate its distinctive properties. Examples suffice to introduce the concept. No doubt this is because it is hard to say anything useful about it (compared to propositional knowledge). This is why I have attempted to get at its general features, acknowledging that it is quite obscure (in what cognitive form is it represented?). It is a lot harder to penetrate than propositional knowledge—we have no convincing analysis of it. Russell says that it is “essentially simpler” than what he calls “knowledge of truths”: that may be so, but it doesn’t follow that it is easier to understand.