Notebook errors

Use waterproof ink Never pencil Pencil will disappear with time, and so will your grade. Cheap ink will wash away and carry your grades down the drain. Never erase Just draw one line through yuor orrcro your errors so that they can still be seen. And never, never, never cut any pages out of the notebook ... 

Note any unusual occurrences, even if they indicate some laziness or inattentiveness on your part. Despite aU earnest attempts to keep an experiment from going awry, the complexity of most experiments conspires to introduce error or accidents into all observations. (It s far better to acknowledge an erroneous or suspicious measurement than to send later readers (including yourself) off on a series of experiments to follow up on a misleading observation that appears in your notebook without comment.)... 

The first course of action would be for the analyst to inspect his/her technique, chemicals, notebook records, and perhaps the equipment used, to try to detect a determinate error. If a cause is found, then the measurement should be rejected and the reasons for such rejection documented. If a cause is not found, and if time is not a factor, it would be advisable to repeat the measurement, perhaps many times, to see if the anomaly appears again. If it does, the situation is not resolved unless a cause is established in the course of the repetition. If it does not appear again, its seriousness has diminished because there are more measurements from which the mean is calculated. 

Calculate the equivalent of zinc. Determine the magnitude of the error in computing it in per cent. Enter the results of your calculations and observations into your laboratory notebook as in the preceding experiment, using Form 5. Proceeding from the known valence of magnesium and zinc and your results, calculate the atomic masses of these elements. Compare the values obtained with tabulated data. 

The critical functions of your lab notebook are to state what you did and what you observed, and it should be understandable by a stranger. The greatest error, made even by experienced scientists, is writing incomplete or unintelligible notebooks. Using complete sentences is an excellent way to prevent incomplete descriptions. 

Each time an error is made, it is initialed, dated, and one of the code numbers in the list is placed next to the initials and circled. A ccpy of the list is placed in the front of each notebook for reference. Pencil or white out are not to be used under ary circumstances. Note at the bottom of Figure 11 the place for the witness or supervisor to sign. During audits, we have had many discussions about this. The consensus in the Office of Compliance Monitoring is, if there is a place for a signature, sign it. If this practice is not acceptable to the laboratory, an SOP should be developed to explain this deviation in the use of the form. 

Whatever style of notebook is used, the principle is the same Record all data directly in your notebook. Data may be copied into tlie notebook from a partner s notebook in those cases where it is clearly impossible for both partners to record data at the same time. Even then, a carbon copy or a photocopy of the original pages is better, since it avoids copying errors and saves time. In particular, do not use odd scraps of paper to record such incidental data as weights, barometer readings, and temperatures with the idea of copying them into the notebook at a later time. If anything must be copied from another source (calibration chart, reference book, etc.), identify it with an appropriate reference. 

A sample calculation should also be presented as an appendix to an undergraduate laboratory report. This appendix should show how one obtains tlie final results starting from the raw data. In general, the numbers used in the computations should have more significant fignres than are justified by the precision of the final result, in order to avoid mathematical errors due to roundoff. Units should be included with each step of the calculation. Also specify the source of raw data used (e.g., mn 5 on page 14 of notebook). 

Remember that you can discard an experimental measurement only if you have certain knowledge that you made an experimental error. Thus, you must carefully record experimental observations in your notebook as soon as they occur. 

Most personal errors can be minimized by care and self-discipline. It is a good habit to check instrument readings, notebook entries, and calculations systematically. Errors due to limitations of the experimenter can usually be avoided by carefully choosing the analytical method. 

Reexamine carefully all data relating to the outlying result to see if a gross error could have affected its value. This recommendation demands a properly kept laboratory notebook containing careful notations of all observations (see Section 21). 

The above example is an abbreAdated version in which actual calculation or numerical setups are omitted. For complete record keeping, you should include the computational setups m your notebook so an error can be tracked down later, if necessary. 

If the anomalous result can be unequivocally assigned to laboratory error, the result may be invalidated. Marking the notebook entry as invalid and retaining all related instrument outputs will be invaluable during future audits, to account for the raw data and results which are retained in the instrument electronic database. 

I have benefited over the years from many discussions with the authors of the papers cited, and those of many other papers which time and space preclude. I thank them for their explanations and inspiration. I also thank Dr Gillian Bade for a critical reading of the manuscript. Any omissions (and I know that there are many) and errors are mine and I apologize to those whose work I may have seemed to overlook. This chapter is not really a history of science. If it were to be so, I would have interviewed many of the protagonists involved in the research, and requested copies of their correspondence, lab notebooks and notes. Nor is it a comprehensive review of the literature which is, at this point, too vast and diversified to encapsulate into a single short article. Rather it is my recollection of the important events which subsequently shaped the field, and my own opinion about what constitutes important. I have attempted to cite the earliest, most seminal papers and to include appropriate and accessible reviews. Finally, I wish to thank Alan Heeger, my thesis advisor until 1975, and a mentor ever since. 

Blunders are one-time mistakes that are random but not normally distributed. They might be the result of misreading an instrument or an incorrect notebook entry. The most egregious blunders can be identified and rejected from the data set using Chauvenet s criterion. Chauvenet s criterion is based on the idea that a typical set of experimental data will have a normal distribution of errors (Taylor, 1982). That is, 68.3% of the data will fall within one standard deviation of the mean 95.5% will fall within two standard deviations and 99.7% will fall within three standard deviations. If the data set contains values that do not conform to this pattern, then these values are likely to be the result of some error(s) other than the expected random errors that produce a normal distribution. Because these unusual values are not part of the expected pattern they should be discarded from the data set. Keep in mind that this approach assumes a normal distribution of error. If the measurement produces a different kind of error distribution, Chauvenet s criterion cannot be used. 

---