Even harder than bringing a locomotive down the 4% light, is bringing it down with a train that has too little braking power for its weight. As someone pointed out before, you can't use the engine's driver brakes much, or you can overheat, and "slip" a tire. All you normally have to use are the tender brakes and the brakes on the cars behind you, which have to hold back not only their own weight, but most of the weight of the engine, as well.
But if you don't have enough braking power in the string of cars to hold you back, you have to use a combination of compression breaking and train brakes. The steam you use for this is saturated steam from the "drifting throttle", so it's at a lower temperature and wetter than superheated steam, but even then you can't run in compression too long, or you'll wash all the lubrication out of the cylinders, and damage them. So you have to switch back and forth between assisting the train brakes with compression, and cooling the cylinders while drifting forward using only the train brakes. Since the grade increases and decreases at various locations, you would use the lesser grades for opportunities to cool the cylinders.
My first experience using compression was as a fireman under Sid McKinney in the fall of 1970. We had a train consisting of about 20 empty cars, with a dead locomotive spliced in the middle. It was raining and getting dark as Sid carefully brought the train down around Windy Point to Coxo. As we were coming out of Coxo, he had me try my hand at it. Coxo is flat enough and a sharp enough curve that the locomotive was drifing in forward motion, even giving the train brakes a little time to recharge. As you pass Coxo siding, the grade steepens, and the train began to pick up speed. Sid told me to put the Johnson bar in reverse and cut in the drifting throttle as I had seen him do. I did, and gradually notched the Johnson bar back to keep the speed in check. Suddenly, the noise of the exhaust and clatter of the running gear became eerily silent. I was puzzled and turned to look at Sid and ask him what was happening, but before I could, he started yelling, "Horse her over, horse her over!" At the same time he lunged for the Johnson bar, and shoved it forward, while turning on the sander valves.
The noise returned and the engine surged forward, as the drivers resumed turning again. Then he quickly reversed the Johnson bar again, increasing it little by little, until the train was descending at a comfortable speed again.
What had happened was that running in reverse on rain-slicked rails had caused the drivers to stop turning. A few more seconds of that, and we would have flattened them. But Sid's reaction was quick enough that no damage was done. One of the strongest impressions that experience made in my mind was of how rapidly the train gained speed during those few moments, and how long it took to slow it back down.