IPv4-64 Corrected Header and Unified Routing Specification: A 36-byte fixed header that routes native IPv4-64, carries legacy IPv4, and transits IPv6 across one forwarding plane

IPv4-64 extends IPv4 addressing from 32 to 64 bits while keeping dotted-decimal notation, a fixed header with no options, and full backward compatibility with every existing IPv4 address. It adopts the IPv6 improvements that proved necessary — a flow label, no per-hop header checksum, and a mandatory UDP checksum — and adds in-header security: source validation, fragment authentication, stateless SYN-flood defense, amplification throttling, and a strict silent-drop rule. It exists for the networks that do not want, need, or can justify IPv6, offering a larger address space with minimal change rather than a parallel protocol stack. This document corrects and extends the original specification. The IP header, previously stated as 32 bytes, could not hold its own fields at their stated widths; it is corrected to a verified 36-byte layout, still 4 bytes smaller than IPv6. A new 4-bit transit_mode field lets one fixed header do three jobs on a single forwarding plane: route native IPv4-64, carry legacy IPv4 at the edges, and transit IPv6 at the /64-prefix granularity using standard protocol-41 encapsulation, with the fabric treated as a single IPv6 hop so the core's internal structure stays invisible. All three modes forward identically — a 64-bit longest-prefix match on one destination field — so the new capability costs one new field and three reinterpretations of existing fields, not a new header.