Forgive my ignorance, date was another field.
Then I would construct the index as :

emp_num + dtos(date) + padl(start_time,5,"0")

Which would make the filter fully optimizable :
set filter to emp_num + dtos(date) + padl(start_time,5,"0") = 'xxxxxx' + dtos({xx/xx/xx})
with set exact off. Even to make it faster, instead of "set filter" :
set order to tag newIndex && newIndex - emp_num + dtos(date) + padl(start_time,5,"0")
set key to 'xxxxxx' + dtos({xx/xx/xx})
Cetin

>The start_time field is just a numeric field that represents the time in 24-hour decimal-hour format.
>
>For instace:
>
>8:15 a.m. is stored as 8.25
>
>
>
>>>I have a TIME CLOCK app in VFP6 using free tables.
>>>
>>>the table structure is
>>>
>>>emp_num (c6)
>>>start_time (N5.2)
>>>stop_time (N5.2)
>>>date (Date)
>>>
>>>I have an index on emp_num and date.
>>>
>>>In the app, I do this:
>>>
>>> use time_clock order emp_num
>>> set filter to emp_num='xxxxxx' and date={xx/xx/xx}
>>>
>>>to see a particular employees time records for a particular day.
>>>
>>>Usually, the records are entered sequentially, such that the order of the records is automatically in chronological order with respect to the "start_time".
>>>
>>>However, sometimes records will be added so that they are NOT in order by "start_time" as entered.
>>>
>>>So, I modified the emp_num index tag be "emp_num+str(start_time,5,2)"
>>>
>>>
>>>Now then, is this a good way? Please advise better method (using free tables).
>>Change your index expression to :
>>if starttime is date
>>

emp_num+dtos(starttime)

>>if starttime is datetime
>>

emp_num+ttoc(starttime,1)

Also you could add "descending" keyword if wat to see last entries first.
>>Cetin