Scientists generally keep notes in a bound, blank book. Some of these books are ruled both directions so each page can be used as graph paper. Write and sketch anything and everything of scientific interest. Write down any questions that may be interesting regardless of whether the idea seems workable now. Someday you, or someone else, may want to look into it. Your notebook is your long-term memory. Professional research scientists are required to maintain laboratory notebooks with all their workday thoughts in them.

The notebook is the scientist's pride and joy. It represents his or her work. It is the scientist's central tool. It is not excessively neat, just useful. In it you will find:

In the context of measurement, accuracy means that the measurement of some quantity is closely associated with the actual quantity. The two are never exactly the same because of a variety of types of "errors." While accuracy is always a goal, precision is only sometimes worth the effort; this is the topic of the section entitled "Estimation" below. Error refers to the imperfections present in all measurements. Sometimes the measured quantity is not clearly defined itself, e.g., the width of a pillow. Sometimes the measured value fluctuates while you are reading it, e.g., the ping-pong ball windspeed indicator discussed on page 47; this kind of error is called "noise." The resolution of the measuring instrument accounts for some of the error, especially when the measured value is "small," e.g., try using a protractor and a plumb bob to measure the angle of slope of a highway.

Good scientists always record the measurement and its corresponding error. For example, suppose you are measuring a field. You could just look at the field and guess that it is about the length of a football field, 100 yards. The error in this measurement is probably as much as 50 yards depending on how practiced you are; you would record the "measurement" as 100 +/- 50 yards or 300 +/- 150 feet. If instead you "pace off" sixty four paces of five and one half feet per pace, you get 64 * 5.5 = 352 feet. Your estimated error may be six inches per pace for a total of 384 inches (384 inches = 32 feet). Record this measurement as 352 +/- 32 feet. Finally, you may use a steel surveyor's tape and measure the field as "exactly" 368.15 feet (122.72 yards). This measurement is not, however, exact; the measured value depends on how tightly you pull the tape. By pulling again and re-measuring you may get 368.19 feet. Report this measurement as 368.17 +/- 0.02 feet (122.723 +/- 0.007 yards).

On any kind of analog scale, always guess to the nearest fourth or fifth or even tenth of the smallest division. If a ruler is marked to the nearest ¼ inch, you should estimate the actual value to the nearest ¼ of ¼ of an inch, i.e., 1/16 of an inch. If a temperature scale is marked in whole degrees, you should estimate to the nearest one fifth of a degree, i.e., 0.2 degrees. If the divisions are widely spaced enough, you might even try guessing to the nearest 0.1 degree.

For the "most scientific" results, measure a thing several times. Average these values together to obtain the "mean. " This is usually considered the best value for that measurement. Standard scientific reporting also requires a report of the "standard deviation." With a calculator or computer spreadsheet program the mean and standard deviation are readily calculated. Scientific calculators discuss these topics in the "statistics" section of the owner's manual. As an example, suppose you measure a distance several times and obtain 3.88 cm, 3.85 cm, 3.87 cm, and 3.82 cm. The mean and standard deviation for these numbers are 3.855 cm and 0.026 cm, respectively. The value would be reported as 3.855 +/- .026 cm.

Before leaving the section on precision, never forget that precision is useless if the "big" part of a measurement is wrong. Suppose for example, that someone measure a car-sized object with a meter stick and reports the length to be 2.855 meters, reported to the nearest millimeter. However, if they were using a one meter stick and miscounted the number of times they had put the meter stick end-to-end so that the actual length was 3.855 meters, then all that precision was a complete waste!