I have used GeoDa to create a .gal weights file from a shape file, Queen congruity. In GeoDa there are no disconnected polygons. When I read this into Jupyter Notebook I am informed that there are 6 disconnected components:

w = libpysal.io.open("data.gal").read()
w.transform='r'

C:Program FilesAnaconda3envsmarkov_envLibsite-packageslibpysalioiohandlersgal.py:185: UserWarning: The weights matrix is not fully connected: 
 There are 6 disconnected components.
  w = W(neighbors, id_order=ids)

However when I get the histogram of connections I see that there are no zero bins. Also the sum of the frequencies, 35672, is exactly the number of areas I have in my shape file.

w.histogram

[(1, 123),
 (2, 734),
 (3, 2446),
 (4, 5596),
 (5, 8155),
 (6, 7722),
 (7, 4971),
 (8, 2731),
 (9, 1397),
 (10, 737),
 (11, 433),
 (12, 238),
 (13, 140),
 (14, 83),
 (15, 59),
 (16, 41),
 (17, 24),
 (18, 13),
 (19, 9),
 (20, 8),
 (21, 6),
 (22, 2),
 (23, 2),
 (24, 1),
 (25, 1)]

Is there anyway to identify the 6 components that it believes to be not connected?