lower cutoff = -2.00°C
Upper cutoff = 2.00°C
Since the mean is 0 and the standard deviation is 1, your sketch should look EXACTLY like the normal standard deviation curve that you should see in many places online and in text books involving statistics. Now to see where the cutoff points are, look at a standard normal table and look for the value corresponding to your 2.3%. The exact value depends on the table. For instance if it's a "Cumulative from the mean" it will give you the area of the curve from the median point to the desired standard deviation. In which case, you'll see values ranging from [0.0, 0.5). If that's the type of table, convert your 2.3% to 0.5 - 0.023 = 0.477, which if you look up the standard deviation for that value will return you the value 2.00 which means that only 2.3% of the values will lie outside of 2 standard deviations. Since the normal distribution curve is symmetrical, that 2.3% figure will apply on both sides of the mean. E.g. 2.3% of the samples will be above 2 standard deviations above the mean, and 2.5% of the samples will be lower than 2 standard deviations below the mean. Meaning that 4.6% of the overall samples will be further than 2 standard deviations from the mean.
In any case, we know that we're rejecting thermometers that are more than 2 standard deviations from the mean. And since our standard deviation is 1.00°C, that means that 2*1.00°C = 2.00°C is how far off the thermometer can read off before getting rejected. So the lower cutoff temperature is -2.00°C and the upper cutoff temperature is 2.00°C.
Your sketch should look like a completely normal standard deviation curve and the lower cutoff should be labeled -2.00°C and the upper cutoff 2.00°C where those lines are at the standard deviation values of -2 and 2. If you wish to indicate the set of thermometers being rejected, then shade those regions below -2 deviations and above 2 deviations.
