section. We also assume the bar is in standard position with the x-axis along the axis of the bar and the ends at x = 0 and x = 1. Let u = u(x, t) be the temperature of the bar where x is position and t is time, and let a2 be the thermal diffusivity constant of the bar. The temperature of the bar is determined by the heat conduction equation a 2u = ut for 0 0, and the side conditions. The side conditions we shall consider are the initial condition u(x, 0) = b for 0 0. In other words, the bar is heated to a uniform temperature b; at time t = 0 the left end is changed to a temperature of 0 while the right end is changed to temperature a, and these end temperatures are then maintained. To simplify the discussion we shall assume a > 0 and 0 < b < a/2. The solution to this problem may be found in many texts such as [1] or [2]. Clearly the steady-state temperature distribution is limt u(x, t) = (a/l) x. If the temperature of the bar is plotted as a function of x for different times t, then a reasonable guess as to the intermediate time behavior of the solution is shown in FIGURE 1. Surprisingly, this behavior only occurs in the special case where b = a/2. If 0 < b < a/2, then for any x where 0 < x < bl/a (where the initial temperature is greater than the steady-state temperature), substitution of large values of t into the solution for that x results in temperatures lower than the steady-state value. For 0 < b < a/2, even where the initial temperature is greater than steady-state, the temperature of the bar will converge to the steady-state value from below. To verify and understand this behavior we must examine the solution of the problem. The