REDIM Statement Windows/Unix
Declaration statement that changes the space allocated to an array that has already been declared by DIM, preserving as much of the data as possible. Unlike other forms of BASIC this handles memory in contiguous sections. Thus, REDIM might overwrite existing variables in memory if you REDIM a small memory block to a large memory block. Also it does not clean up after each REDIM statement so memory usage keeps growing with each REDIM statement.

Syntax: REDIM variable[(subscripts)] AS type
DIM B(20) AS INTEGER
REDIM B(100) AS INTEGER

Details:
If the array has not yet been declared by DIM, REDIM will be equivalent to calling DIM (except when dealing with components, in which case you'll receive a compiler error). You cannot change the data type for the array, ie. you cannot redefine array B to be something other than an INTEGER (as in the above example). Combining arrays is a by-product of REDIM.
DIM A(1 to 100) AS INTEGER
DIM B(1 to 100) AS INTEGER
REDIM A(1 to 200) AS INTEGER   '-- Combines A and B together
What the above code does is preserve the data in A as well as combining the array B.so index from 101 to 200 is the data for B and 1 to 100 is the data for A. Clarification: Array Bhas not been modified, it still maintains its own address space and will notoverwrite A's data and vice versa. Note that this by-product does not apply to redimming Fixed Strings, Variants and Components.
Redim table
  DIM Button(1 TO 2, 1 TO 2) AS QBUTTON
  REDIM Button(1 TO 2, 1 TO 4) AS QBUTTON
Note that when dealing with multiple dimensions the elements are shifted to fill each column. So element (2,2) is actually moved to position (1,4). 
Some problems to be aware of:
DIM A(1) AS STRING
REDIM A(20) AS STRING    'this redim works fine!

DIM A(1) AS STRING*4
REDIM A(20) AS STRING*4   'this redim causes an error!