Oracle NLS_LANG for language/territory/character set

Recently I had an Oracle problem at work. All database data that had accented characters as à, é, í, ã, etc and other ones as ç (cedilla), º (ordinal indicator) were being replaced by erroneous characters as inverted question marks (¿). The problem was related to NLS_LANG variable that was set on my Windows System variables. To see how to solve this, take a look a this post that shows you the problem, the questioning and the solution.