Empty list bigger in size than corpus object in nltk

If some works with nltk and does a

from nltk.book import *

she will discover that text1 has object size of 64 bytes doing an

import sys
sys.getsizeof(text1)

If now she does a

text1 = []

the size of text1 increases to 72 ! Why is that? Why an empty list occupies more memory than its previous object form in my machine?

Does anyone know something about this? Thank you

via Chebli Mohamed