sql server - Decimal precision and scale -

this expression uses implicit conversion:

select 9.999 * '9.999' 

and evaluates 99.980001.

if log type information using this:

select     sql_variant_property(9.999 * '9.999', 'basetype'),     sql_variant_property(9.999 * '9.999', 'precision'),     sql_variant_property(9.999 * '9.999', 'scale'),     sql_variant_property(9.999 * '9.999', 'maxlength') 

i get:

• basetype numeric
• precision 9
• scale 6
• maxlength 5

i believe understand results except precision. if number of digits left decimal point 2 , right 6, totals 8.

why sql server calculate 9 here?

p.s. use sql server 2008 r2

from precision, scale, , length (transact-sql)

looking @ section

operation: e1 * e2 

you find

result precision p1 + p2 + 1 

so in case 4+4+1 => 9

further that, notice that

select  sql_variant_property(99.999 * '9.999', 'precision'), --11 => p1+p1+1         sql_variant_property(99.999 * '99.999', 'precision'), --11 => p1+p1+1         sql_variant_property(99.999 * 9.999, 'precision'), --10 => p1+p2+1         sql_variant_property(99.999 * 99.999, 'precision') -- 11 => p1+p2+1 

whereas

select  sql_variant_property(9.999 * '9.999', 'precision') --9 => p1+p1+1 

and

select  sql_variant_property(9.99 * '9.999', 'precision') 

cuases

arithmetic overflow error converting varchar data type numeric.