On UNIX-like systems, you should always start scripts with a shebang line. The system call execve (which is responsible for starting programs) relies on an executable having either an executable header or a shebang line.
From FreeBSD’s execve manual page:
The execve() system call transforms the calling process into a new process. The new process is constructed from an ordinary file, whose name is pointed to by path, called the new process file. [...] This file is either an executable object file, or a file of data for an interpreter. [...] An interpreter file begins with a line of the form: #! interpreter [arg] When an interpreter file is execve'd, the system actually execve's the specified interpreter. If the optional arg is specified, it becomes the first argument to the interpreter, and the name of the originally execve'd file becomes the second argument
Similarly from the Linux manual page:
execve() executes the program pointed to by filename. filename must be
either a binary executable, or a script starting with a line of the
form:#! interpreter [optional-arg]
In fact, if a file doesn’t have the right “magic number” in it’s header, (like an ELF header or #!), execve will fail with the ENOEXEC error (again from FreeBSD’s execve manpage):
[ENOEXEC] The new process file has the appropriate access
permission, but has an invalid magic number in its
header.
If the file has executable permissions, but no shebang line but does seem to be a text file, the behaviour depends on the shell that you’re running in.
Most shells seem to start a new instance of themselves and feed it the file, see below.
Since there is no guarantee that the script was actually written for that shell, this can work or fail spectacularly.
From tcsh(1):
On systems which do not understand the `#!' script interpreter conven‐ tion the shell may be compiled to emulate it; see the version shell variable. If so, the shell checks the first line of the file to see if it is of the form `#!interpreter arg ...'. If it is, the shell starts interpreter with the given args and feeds the file to it on standard input.
From FreeBSD’s sh(1):
If the program is not a normal executable file (i.e., if it
does not begin with the “magic number” whose ASCII representation is
“#!”, resulting in an ENOEXEC return value from execve(2)) but appears to
be a text file, the shell will run a new instance of sh to interpret it.
From bash(1):
If this execution fails because the file is not in executable format, and the file is not a directory, it is assumed to be a shell script, a file containing shell commands. A subshell is spawned to execute it.
You cannot always depend on the location of a non-standard program like bash. I’ve seen bash in /usr/bin, /usr/local/bin, /opt/fsf/bin and /opt/gnu/bin to name a few.
So it is generally a good idea to use env;
#!/usr/bin/env bash
If you want your script to be portable, use sh instead of bash.
#!/bin/sh
While standards like POSIX do not guarantee the absolute paths of standard utilities, most UNIX-like systems seem to have sh in /bin and env in /usr/bin.