integer instead of 'double precision'?

take any table and run

Query
---------------------
select
1/3
from
storage
limit 1

Result
---------------------
?column?
integer
0

Expected Result
---------------------
?column?
double precision
0.33333...

Question
---------------------
Since there is no column type to begin with as this is a made-up column,
shouldn't postgres know it is double precision due to the remainder?

I thought perhaps I could cast it as double precision as noted on
http://www.postgresql.org/docs/8.3/interactive/sql-expressions.html

though doing the following:

select
float8(1/3)
from
storage
limit 1

results in:

float8
double precision
0

any ideas on how to get this type of a manufactured column (not sure the
right term for it) to show the double precision result?