Running flate-backref-crosses-blocks.flate through script/print-bits.go and adding commentary: offset xoffset ASCII hex binary 000000 0x0000 . 0x00 0b_0000_0000 Non-compressed block (BTYPE=00), non-final (BFINAL=0) 000001 0x0001 . 0x04 0b_0000_0100 Block length is 0x0004 (parity value is 0xFFFB) 000002 0x0002 . 0x00 0b_0000_0000 000003 0x0003 . 0xFB 0b_1111_1011 000004 0x0004 . 0xFF 0b_1111_1111 000005 0x0005 a 0x61 0b_0110_0001 Literal "abcd" 000006 0x0006 b 0x62 0b_0110_0010 000007 0x0007 c 0x63 0b_0110_0011 000008 0x0008 d 0x64 0b_0110_0100 000009 0x0009 . 0x03 0b_0000_0011 Fixed Huffman block (BTYPE=01), final (BFINAL=1), see below 000010 0x000A B 0x42 0b_0100_0010 000011 0x000B . 0x00 0b_0000_0000 The fixed Huffman block consists of a back-reference that refers to the previous (non-compressed) block, and then an "end of block" code. The length=3, distance=2 back-reference decodes to "cdc", so the overall decoding is "abcdcdc". The breakdown of the fixed Huffman block is: 000009 0x0009 . 0x03 0b_...._.011 Fixed Huffman block (BTYPE=01), final (BFINAL=1) 000009 0x0009 . 0x03 0b_0000_0... (lcode = 257) length = 3 000010 0x000A B 0x42 0b_...._..10 000010 0x000A B 0x42 0b_.100_00.. (dcode = 1) distance = 2 000010 0x000A B 0x42 0b_0..._.... (lcode = 256) end of block 000011 0x000B . 0x00 0b_..00_0000 This back-reference is valid, even though there is no prior data in that block, because there is prior data in previous blocks. RFC 1951 section 2 "Compressed representation overview" says that "the LZ77 algorithm may use a reference to a duplicated string occurring in a previous block".