All files in RDB, Relational Data Base, should be represented as "flat files."　The "flat files" can be normalized by removing groups into separate tables, through the "First Normalization" in Codd's notation(*), as shown below:

　　｛item-NO｜item-name, price-1, price-2, price-3｝

　on the assumption that

　(1) The price-1 implies a fixed price (original price),
　(2) The price-2 implies a discount price for customers who buy a bulk of the items in a transaction,
　(3) The price-3 implies a special price for the customers who have done much trade so far and will do for a long time to come.

　So, we can get the normalized form as below:

　　ITEM: ｛item-NO｜item-name｝ [ R ]
　　ITEM. PRICE:｛item-NO(R)｜assortment-code, price｝[ ER ]

foot-note:

□ The "repeating group" (of attributes) are said to be "polysemic" data in T-ERD.

□ The sign of "ER" on the right side of "ITEM. PRICE" tuple means "Entity. Role."
　"Role" implies the "polysemic" data of attributes in an entity.

□ The "assortment-code" is newly required to assort the prices, such as with the value of "1" (in the assortment-code) for the original price, "2" for the discount price and "3" for the special price. So, you can identify the prices respectively.

□ The "assortment-code" is essentially different from the "partition-code".　The "partition-code" is the code for creating subsets, and the subsets can stand as data-clusters (in a set) at a time, although the prices cannot because the prices are "polysemic" data of an attribute.
　In other words, the "assortment-code" is applied for sorting data, multiple data of a "polysemic" attribute of an entity, therefore "entity. role" never has the identifier because it cannot be an entity.