A packet of twelve cards is introduced which consists of mixed pairs of six cards. The object of the experiment is for the magician to correctly predict which two cards will match up after the spectator spells out the phrase “Magic Trix Are Great Fun” in whichever manner they choose.
The cards are separated into two six-card packets and the spectator has a free choice as to which packet they will begin to spell from. Whichever packet is selected the spectator is instructed to remove one card at a time from the face of the packet and transfer the card to the bottom of the packet for each letter of the word being spelled. Whenever a word is completely spelled then whatever two cards are now on the face of each packet are removed and placed aside and are considered a pair. This is continued until all the words are spelled and there are six pairs of cards face up on the table.
What has not as yet been mentioned (and what makes this effect so baffling) is that the spectator has the option of stopping the spelling of a word with one packet and continuing the spelling with the other packet at any point they desire. They can stop the spelling midway through the word and finish spelling that word with the other packet and then jump back and forth between the packets as they wish. It would seem impossible that the magician could predict the actions of the spectator or the outcome of the paired cards. Yet after the entire phrase “Magic Trix Are Great Fun” has been spelled out, card by card, there will be one pair of cards that match, and those will be the two cards the magician has predicted!
Now for the killer ending to this effect. Even though the remaining five pairs of cards do not match on the face, when they are turned over, they are found to be paired up with their backs of which there are five different types! None of these different back designs match the predicted pair which is different from all the rest!