I've gone for simplicity, and abandoned any attempt at making sense of the data's internal structure. Its just a Block of float values this way

CREATE TABLE BlockRef (

pKey int identity(1,1) PRIMARY KEY,

FileNo int not null,

BlockNo int not null,

itemCount int not null )

CREATE TABLE Block (

pKey int identity(1,1),

fk_BlockRef int not null FOREIGN KEY REFERENCES BlockRef(pKey),

sequence int not null -- sequence of the float value in the original file block,

fValue float )

-- Then to select an array

CREATE PROC GetDataBlock

@FileNo int,

@BlockNo int

AS

SELECT fValue

FROM Block

WHERE fk_BlockRef = (

SELECT pKey FROM BlockRef

WHERE FileNo = @FileNo

AND BlockNo = @BlockNo )

ORDER BY sequence

I think this will work and I've started on the C program to populate the data structure

I'll put some indexes on it too.