If we compressed our data using Huffman coding, but had to send this code word dictionary along with the compressed data, then we obviously only get an advantage if the Huffman coded data saves us at least as many bits as were going to have to add to define the dictionary. If Huffman coding generated a file (without the code word dictionary) that is 25% smaller than the original, how big would the original text file have to be before it was worth Huffman coding, if we wanted to send the code word dictionary along with the compressed data?