You cound try something like:

create table #junk (

float_to_add float NOT NULL)

go

insert #junk(float_to_add) values (1.1104)

insert #junk(float_to_add) values (1.222)

insert #junk(float_to_add) values (1.32)

insert #junk(float_to_add) values (1.41)

insert #junk(float_to_add) values (1.52)

insert #junk(float_to_add) values (1.66)

insert #junk(float_to_add) values (1.71)

go

select float_to_add from #junk

go

select sum(convert(numeric(13,3),float_to_add) % 1.000) from #junk

go

drop table #junk

go

But float is not an exact datatype, it's an approximation, so you may get some unexpected results because a number like 1.3 might actually come out as 1.29999999999.

If the numbers you're storing need definite precision, float may not be the appropriate data type.