To usefully distinguish particulars from universals, Russell posits the example of "the most long-lived of men," a description which wholly consists of universals. We assume that the description must apply to some man, but we have no way of inferring any judgment about him. Russell remarks, "all knowledge of truths, as we shall show, demands acquaintance with things which are of an essentially different character from sense-data, the things which are sometimes called 'abstract ideas', but which we shall call 'universals'." The description composed only of universals gives no knowledge by acquaintance with which we might anchor an inference about the longest-lived man. A further statement about Bismarck, like "The first Chancellor of the German Empire was an astute diplomatist," is a statement that contains particulars and asserts a judgment that we can only make in virtue of some acquaintance (like something heard or read).
Statements about things known by description function in our language as statements about the "actual thing described;" that is, we intend to refer to that thing. We intend to say something with the direct authority that only Bismarck himself could have when he makes a statement about himself, something with which he has direct acquaintance. Yet, there is a spectrum of removal from acquaintance with the relevant particulars: from Bismarck himself, "there is Bismarck to people who knew him; Bismarck to those who only know of him through history" and at a far end of the spectrum "the longest lived of men." At the latter end, we can only make propositions that are logically deducible from universals, and at the former end, we come as close as possible to direct acquaintance and can make many propositions identifying the actual object. It is now clear how knowledge gained by description is reducible to knowledge by acquaintance. Russell calls this observation his fundamental principle in the study of "propositions containing descriptions": "Every proposition which we can understand must be composed wholly of constituents with which we are acquainted."
Indirect knowledge of some particulars seems necessary if we are to expressively attach meanings to the words we commonly use. When we say something referring to Julius Caesar, we clearly have no direct acquaintance with the man. Rather, we are thinking of such descriptions as "the man who was assassinated on the Ides of March" or "the founder of the Roman Empire." Since we have no way of being directly acquainted with Julius Caesar, our knowledge by description allows us to gain knowledge of "things which we have never experienced." It allows us to overstep the boundaries of our private, immediate experiences and engage a public knowledge and public language.
This knowledge by acquaintance and knowledge by description theory was a famous epistemological problem-solver for Russell. Its innovative character allowed him to shift to his moderate realism, a realism ruled by a more definite categorization of objects. It is a theory of knowledge that considers our practice of language to be meaningful and worthy of detailed analysis. Russell contemplates how we construct a sense of meaning about objects remote from our experience. The realm of acquaintance offers the most secure references for our understanding of the world. Knowledge by description allows us to draw inferences from our realm of acquaintance but leaves us in a more vulnerable position. Since knowledge by description also depends on truths, we are prone to error about our descriptive knowledge if we are somehow mistaken about a proposition that we have taken to be true.
Critics of this theory have held that Russell's hypothesis of knowledge by description is confusing. His comments when defining sense-data, that the physical world is unknowable to us, contradict his theory of knowledge by descriptions. He implies that "knowledge by description" is not really a form of knowledge since we can only know those things with which we are acquainted and we cannot be acquainted with physical objects. Russell's theory amounts to the proposition that our acquaintance with mental objects appears related in a distant way to physical objects and renders us obliquely acquainted with the physical world. Sense-data are our subjective representations of the external world, and they negotiate this indirect contact.
While innovative, Russell's theory of knowledge by description is not an attractive theory of knowledge. It is clearly unappealing because our impressions of the real world, on his view, are commensurate with muddy representations of reality. Though we have direct access to these representations, it seems impossible to have any kind of direct experience of reality. Reality, rather, consists in unconscious, inferential pieces of reasoning.