Data anomalies

A friend working in the medical data analysis business explained to me that nobody every has diastolic blood pressure of 89, 90, or 91. This is because 90 (mmHg) is the threshold for high blood pressure, so when a doctor or nurse measures your blood pressure, if it’s 89 or 90 or 91 but you are otherwise healthy, they squint and write down 88. That way they and you don’t have to deal with all of the extra paperwork and hassle of having high blood pressure. It’s only when your blood pressure is 92 or 93 or higher that they decide that it’s in your best interest to be recognized as having high blood pressure. So if you look at aggregate blood pressure data, there’s a gap between 88 and 92.

The Wall Street Journal reports on China's preparations for the Olympics, and includes Beijing's pollution index for the last year. The numeric rating corresponds to concentrations of various pollutants in the air. China considers a rating above 100 to be dangerous; assuming the scale is consistent with that used in other countries, that's two to three times the level that triggers a warning elsewhere. Even so, there's something peculiar about this data:

In exactly which counting system is it normal to round numbers between 51 and 70 down to 50, and 101 and 120 down to 100, but leave all other numbers apparently untouched?