You really should place new questions in new threads, if for no other reason than more people are likely to see it.
For this problem it looks like you've got the right ideas, but you're losing way too much accuracy by not keeping enough decimal places in your intermediate results. Keep intermediate values in fraction form or keep several more decimal places ("guard digits").
For example, knowing that I1 is 4 mA and it's flowing through a 4 kΩ resistor you can tell that the potential drop across that resistor should be 16 V exactly. That means the sum of the 42V supply and the drop across the 6 kΩ resistor must add to 16V, so that the drop is 26 V across that resistor. But in your determination of i2 you rounded the result to 4.3 mA, and then when you calculated the potential drop across that resistor you arrived at 25.8 V.
The situation gets worse when you round small values, as the resulting percentage change is more drastic. You determined the current through the 18 kΩ resistor to be 1.63 mA, but it should have ended up as 1.889 mA. So the early roundings of values has landed you with a 14% error so far...