T-formed ERD uses a square to denote a data grouping and a vertical line within the square:
　its format is so "T-formed"--the same format of a journal where accountants report transaction data--;
　the left side is the place where the "identifier" is stored, and the right where the "attributes" are stored.

　
● Data clustering

　A single entity (occurrence) must have an identifier. Most computer systems assign an arbitrary number or code as the identifier: employee-number, division-code, order-number etc.
　The identifier must NOT uniquely identify a single occurrence: the uniqueness for accessing data through indexing with the (master) key shall not be now mixed up with the identifier, which is recommended later in the process of creating indexes.

　Attributes are data elements containing a single piece of information about entity: i.e.,
　　　name, date, amount

　Attributes are, therefore, named under the rule of "entity. attribute": i. e.,
　　　employee. name　　　　order. date
　　　customer. name　　　　delivery. date
　　　product. name　　　　　billing. date

　Take an attribute, and just ask "of which?": i. e.,
　attribute of "name", and ask "whose?" employee's? customer's? product's?
　attribute of "date", and ask "the date of which?" order? delivery? billing?

NOTE:
　Most COBOL programs suffer from a proliferation of homonyms, aliases, and synonyms.　Studies have indicated that there are 20 "other" names for each real data element.
□