The Library of Congress classification system divides all knowledge into twenty-one basic classes, each identified by a single letter of the alphabet. Most of these alphabetical classes are further divided into more specific subclasses, identified by two-letter, or occasionally three-letter, combinations. For example, class N, Art, has subclasses NA, Architecture; NB, Sculpture, ND, Painting; as well as several other subclasses. Each subclass includes a loosely hierarchical arrangement of the topics pertinent to the subclass, going from the general to the more specific. Individual topics are often broken down by specific places, time periods, or bibliographic forms (such as periodicals, biographies, etc.). Each topic (often referred to as a caption) is assigned a single number or a span of numbers. Whole numbers used in LCC may range from one to four digits in length, and may be further extended by the use of decimal numbers. Some subtopics appear in alphabetical, rather than hierarchical, lists and are represented by decimal numbers that combine a letter of the alphabet with a numeral , e.g. .B72 or .K535. Relationships among topics in LCC are shown not by the numbers that are assigned to them, but by indenting subtopics under the larger topics that they are a part of, much like an outline. In this respect, it is different from more strictly hierarchical classification systems, such as the Dewey Decimal Classification, where hierarchical relationships among topics are shown by numbers that can be continuously subdivided.
From the Library of Congress https://www.loc.gov/catdir/cpso/lcco/