You are on page 1of 4
Pet Cte Errors in fe eetec@ atlas «stati sitions discussed in this chapter are fundamental tothe use ol tats fgg treaty data an for testing various hypotheses, Th quincune shown in the ope pao ie echanical device tha forms anormal probaity ition, Every 10 minutes, 30.000 bal op tome center tp of the machine, whch contains a regular pattem of pegs to randomly dati the balls Eochtine 2 bal its a ex ithas a 0.50 chance of fling tthe right r othe. Aer each al pass treh the aay of pegs it drops into oneal the vertical “ins” of he anspaen ase. The umn of basin each bin is proportional othe probability ofa ball falling into given bi, The sn ‘hove shown in the boto half ofthe phot traces out the probabiity distribution ‘andom errors are present in every measurement no matter how careful the experimenter. In this chapter, we consider the sources of random errors, the determination of theit magnitude, and their effects on computed results of chemical analyses. We also introduce the significant figure convention and ilustrate its use in reporting analytical results (Gil THE NATURE OF RANDOM ERRORS Random, or indeterminate, errors can never be totally climinated and are often in a determination. Random errors are caused jor source of uncertaint by the many uncontrollable variables that accompany every measurement, Ust ally, most contributors to random error cannot be positively identified. Even if we can identify random error sources, it is often impossible to measure them cannot be detected individually. The accu- because most are so small that th mulated effect of the individual uncertainties, however, causes replicate results to fluctuate randomly around the mean of the set. For example, the scatter of data in Figures 5-1 and 5-3 is a direct result of the accumulation of small random uncertainties. We have replorted che Kjeldahl nitrogen data from Figure 5-3 as a three-dimensional plot in Figure 6-1 to better reveal the precision and accu- racy of each analyst. Note that the random error in the results of analysts 2 and 4 is much larger than that seen in the results of analysts 1 and 3. The results of analyst 3 show outstanding precision but poor accuracy. The results of analyst 1 show excellent precision and good accuracy. Scanned with CamScanner 1 are both previse and accurate ts of analyst 3 are precise 8 three-dimensinal plot hate cere in Kjeldahl foe tour aitfer that the results of cermor is lange. The results and 4 are both imprecise m Error Sources ieative idea of the way small undetectable iv y le uncertainties We ean get a quali he fall ie Me a om err in he allowing wa. Imaging situation in d F Which jue ive an overall error. We will assu ich jus go assume fous thar exe all random errors combine hac an equal probability of occurring and that each can cause the fi, eng al reg high or low by a fixed amount 2. he possible ways the four errors can co; ‘an combi: NE 10 Bive “Table 6-1 shows all ¢ indicated deviations from the mean value. Note that only one combi, nati be ation ede TABLE 6-1 | Number of | ‘Combinations of ‘Magnitude of _Uecrainies__Random Ervor_Combaons _Eguag, +; + U, + Uj + Us +4U T =U, 4 U; + Ut Us [aunt urn Jt Ut - + Uy +2U 4 (4u+G4+U-U, ns 035 — UU, + U4 Us t+U,+U,-U- Uy +U- Ut Uy -U,+U,-UtU, 0 6 -U+U+tU-\ 6/16 = 0.375, HU, -U;,- Ut 6A-1 Rando! FU, -U;- Us— U; ~U+U-U-Y TY - V+ U- U, -wW -U-U- Ut, 4 41NG = 0.250 1/16 = 0.0625 Yi ii DM aes deems SEM Scanned with CamScanner GA The Nature of Random Errors 98, jon of +4 U, four com oF O Us The 8 give a dew cpative devia crtors have ofs2U ave af he probably Je same tel large number of measuremen E darthews in\Figurt @2n Nene ae Figure 6-2a, Note thatthe ache eek tue it possible combinations nt Bat the reaive Convene exe ll she uncertainties have the same ‘magnitude. This restriction In not necessary to derive the equation for a Gaussian curve ye see that the most frequent occurrence is zero desi vigher extreme a maximum devia sri uncer ‘When the same procedure jon of 10 V occurs on ion rom the mean At : is applied to a v ly about once in 500 res 4 bell-shaped curve like that feos id rery large number orinetadead = spate cre I orn ure 6-2e results, Such » plot i called Relative frequeney cu AU 2U 0 #20 440 +60 ‘Deviation fro mean gA-2 Distribution of Experimental Result: s From experience with many determi yy determinations, we find that the distribution of rep- @ freaee data from most quantitative analytical persian curve shown in Figute 6-2c. As an ample. spproaches that of the te gheet in Table 6-2 for the calibration of ar a Se aa eee srr flask and stopper were weighed. Ten ites Eee 2 ees with the pipet, and the flask was stoppered. The Mask ee ans mine its density. The mass of tine water was then calculated by ‘sking sured to deter sieyifference between the two masses. The mass of water divided by is density is the volume delivered by the pipet. ‘The experiment was repeated The data in Table 6-2 are typical of those obtained ti an ears worker © weighing to the nearest milligram (which corresponds to 0.001 mL) on a top-loading, Falance and being careful to avoid systematic error. Even so, the results vary from 2 ow of 9.969 mL to a high of 9.994 ml. “This 0.025-mL spread of data results direcely from an accumulation of al random uncertainties in the experiment. ‘The information in “Table 6-2 is easier to visualize when the data are rearranged into frequency distribution groups 2s ip Table 6-3. In this instance, we count and tabulate the number of data points falling into a series of adjacent 0.003-mL ranges and calculate the percentage rrr measurements in each range. Note that 2646 of the fhe volume range from 9.981 £0 9.983 mL. This is the group con” ee a vind median value of 9.982 mL. Nore also that more than half thee) a 0.004 mL of this mean. distribution data in “Table 6-3 are plotted as @ bar graph, or oa 6-2 Frequency di * agi for measurements containi histogram in Figure 6-3. labeled A. We can imagine that as the number OF ndom uncertinis, ©) measurements increases, the histogram approaches the shape of the continuous uncertainties nd (© a9 curve shown as plot 3, This plot shows a Gaussian curve, or normal number of random une! rve, which applies © eet of data. The Gaussian CONE The spread in 24 ofre and the sal .¢ differen spre: Relative Frequent 1 . ol -12U -8U -8U_ 0 +40 98U* Deviation from mean Relative frequency results occur in ¢ taining the mean results are within The frequency Bin Figure 6~ oan infinitely larg’ me area under the surements is the error cul ge hee the same mean (9.982 mL the same precision, oy : vighes west FES curve as the histogram. Highest and lowest es [A histogram is a bar g that shown by plot Ai _ TSee Section 38A-4 for an experiment O° calibration of a pipet Scanned with CamScanner 6B-2 Properties of Gaussian Curves Fee 6 fe nos oe Gaussian curves in which we plot the relative frequency y of s the deviation from the mean. As shown in the margin, curves such as these can be described by an equation that contains jus two parameters, the population mean ys and the population standard aie ° The term parameter refers to quantities such as jz and o that define a population or distribution. Data values such as x are variables. The term statistic refers to an estimate of a parameter that is made from a sample of data as discussed below. The sample mean and the sample standard deviation are examples of statistics that esti- mate parameters [L and @ respectively. The Population Mean p. and the Sample Mean x Scientists find it useful to differentiate between the sample mean and the population mean. The sample mean x is the arithmetic average of a limited sam- ple drawn from a population of data. The sample mean is defined as the sum of ded by the number of measurements as given by the measurement values divi N represents the number of measurements Equation 5-1, page 84. In that equation, 0.4 03 > 3 3 3 g g 5 2. & z 2 ba & = 02 ° s 2 Scanned with CamScanner

You might also like