Choose K≥0 with ∣u(s)∣≤K for all s∈[0,T]. Since u is measurable and bounded and [0,T] has finite Lebesgue measure, u is integrable on [0,t] for every t, so the hypothesis is meaningful.
Define v:[0,T]→R by
v(t)=a+c∫[0,t]u(s)ds.
Step 1: v is continuous. Let 0≤r≤t≤T. Writing u1[0,t]=u1[0,r]+u1(r,t], where 1D denotes the function equal to 1 on D and 0 elsewhere, and using linearity of the integral,
v(t)−v(r)=c∫(r,t]u(s)ds.
Since −K≤u≤K, monotonicity of the integral gives ∣v(t)−v(r)∣≤cK(t−r). Hence v is continuous on [0,T]; in particular v is measurable and bounded, so it is integrable on each [0,t].
Step 2: v satisfies the same inequality. By hypothesis u(t)≤v(t) for every t∈[0,T]. Since c≥0, monotonicity of the integral gives, for every t∈[0,T],
v(t)=a+c∫[0,t]u(s)ds≤a+c∫[0,t]v(s)ds.
Step 3: conclusion. The function v is continuous, so its Lebesgue integral over [0,t] coincides with its Riemann integral there, by the compact-interval toolkit. Thus the inequality of Step 2 is exactly the hypothesis of Gronwall's lemma in integral form for the continuous function v with constants a and c≥0, and that lemma yields
v(t)≤aectfor every t∈[0,T].
Combining with u(t)≤v(t) gives u(t)≤aect for every t∈[0,T]. ■