When you record an operation on an object, QuickTest adds 
the appropriate test object to the object repository. After 
the object exists in the object repository, you can add 
statements in the Expert View to perform additional methods 
on that object. To add these statements, you usually enter 
the name (not case sensitive) of each of the objects in the 
object's hierarchy as the object description, and then add 
the appropriate method.But in descriptive programming it is 
quite different.Let us see the below example.

Decriptive programming with an example on a webpage.

General programming:

Browser("Mercury Tours").Page("Mercury Tours").WebEdit
("Name:=Author", "Index:=3").Set "Uma Sankar" 

Descriptive Programming:

Browser("Title:=Mercury Tours").Page("Title:=Mercury 
Tours").WebEdit("Name:=Author", "Index:=3").Set "Uma Sankar"

The difference is quite very less. But is very important.

Note:If you specify a test object by its object repository 
name after other objects in the hierarchy have been 
specified using programmatic descriptions, QuickTest cannot 
identify the object. 

If any doubts please let me know.

wtiting Script with out having the object repository.
or,

 A technique used to handle automation test even though 
object repository is not available.