Unexpected results Min() Max() results from SELECT TOP 100 sql statement

Testing a solution to find min() max() of 100 rows returned from a query provided some interesting results.

SELECT  min(x.radar_queue_key), max(x.radar_queue_key)
FROM (SELECT TOP 100 radar_queue_key FROM z_radar_queue WHERE radar_group_Code='SFV OTS' ORDER BY radar_queue_key asc) as x

Results:867 966

SELECT  min(x.radar_queue_key), max(x.radar_queue_key)
FROM (SELECT TOP 100 radar_queue_key FROM z_radar_queue WHERE radar_group_Code='SFV OTS') as x

Results:867 1879

SELECT  min(radar_queue_key), max(radar_queue_key)
FROM  z_radar_queue 
WHERE radar_group_Code='SFV OTS'

867 1879

What is interesting is that actual Min() =867 and actual max() =966 for the Top 100 rows.

Both of these queries return exactly the same rows in exactly the same order (even when flushing data cache) because radar_queue_key is the primary key on the table.

SELECT TOP 100 radar_queue_key FROM z_radar_queue WHERE radar_group_Code='SFV OTS' ORDER BY radar_queue_key asc

SELECT TOP 100 radar_queue_key FROM z_radar_queue WHERE radar_group_Code='SFV OTS'

Value of 1879 does not appear in either result set. The execution is completely different if the sql statements are used as a subquery.

Lesson here is always use an order by clause when fetching rows for aggregation + TOP xxxx