1

00:00:13,840 --> 00:00:14,620

Okay!

2
00:00:14,860 --> 00:00:23,460
Welcome again, because we are here in basic surveying video lecture and today, we
will

3
00:00:23,470 --> 00:00:31,210
be talking about, again, module number 2 and ours will be lecture number 3 in
module number 2.

4
00:00:32,520 --> 00:00:35,820
The module is ‘Basic Concepts of Surveying’.

5
00:00:37,520 --> 00:00:46,240
Well, we have already done module 1; we are in module 2. In module 2, what we

6
00:00:46,240 --> 00:00:53,880
will be discussing today is on chapter 3,
or rather, the lecture number 3.

7
00:00:55,020 --> 00:00:56,980
In our last video lecture What we did in our last video lecture was very
important, and let us recapitulate that.

8
00:00:57,140 --> 00:01:00,480
We talked about the basic principles of surveying.

9
00:01:01,540 --> 00:01:04,700
This is very important, what are the basic principle of surveying?

10
00:01:06,240 --> 00:01:13,140
well It teach what will be discussing today in chapter 3.

11
00:01:13,820 --> 00:01:20,279
or other than lecture number 3. what we did in our last video lecture was very
important.

12
00:01:21,120 --> 00:01:26,540
and that recaptulate that. we talked about the basic principle of surveying.

13
00:01:27,760 --> 00:01:32,039
These are the principles which we should always keep in our mind whenever we are
going for

14
00:01:32,039 --> 00:01:38,990
surveying because they are very, very important points, and they control the kind
of the work

15
00:01:38,990 --> 00:01:41,610
which you ultimately produce at the end of
your job.

16
00:01:42,740 --> 00:01:46,399
Well, number one is the reconnaissance.

17
00:01:46,399 --> 00:01:52,530
What we did in our last video lecture, we
started with an exercise - an exercise in

18
00:01:52,530 --> 00:01:57,880
which we started - well, there is a little
ground, and we want to make a map of that

19
00:01:57,880 --> 00:01:58,880
ground.

20
00:01:58,880 --> 00:02:01,420
So what all steps are involved?

21
00:02:01,420 --> 00:02:07,039
The very first one was recy survey or reconnaissance
- we go to the ground, observe it - this is

22
00:02:07,039 --> 00:02:09,910
very important point in any surveying job.

23
00:02:09,910 --> 00:02:16,030
Then the second one; we decided ‘Okay, we
are going to survey this particular area by

24
00:02:16,030 --> 00:02:18,250
a particular method’.

25
00:02:18,250 --> 00:02:22,860
In our last video lecture we chose the method
of triangulation.

26
00:02:22,860 --> 00:02:28,510
For this triangulation, we need some points
in that area so that we can form the triangle.
27
00:02:28,510 --> 00:02:31,840
Now, why we are forming the triangle?

28
00:02:31,840 --> 00:02:37,239
Because we wanted to bring the skeleton of the area in our plotting sheet.

29
00:02:37,239 --> 00:02:42,480
So, that idea is called ‘working from whole
to part’ or ‘establishing the control

30
00:02:42,480 --> 00:02:43,480
network’.

31
00:02:43,480 --> 00:02:48,780
We want to establish first a very accurate
control network, and we will be using that

32
00:02:48,780 --> 00:02:51,900
control network later on for our other surveys.

33
00:02:52,560 --> 00:02:58,180
So, we do not work from part to whole, rather, we work from whole to part.

34
00:02:58,180 --> 00:03:03,300
First we make the entire skeleton; we are
not bothered about the details initially.

35
00:03:03,310 --> 00:03:08,860
What we are bothered about initially is the
skeleton - it has to be accurate, and the

36
00:03:08,860 --> 00:03:10,660
details can be filled in later on.

37
00:03:11,680 --> 00:03:16,180
Then another thing - we must always go for
the redundancy.

38
00:03:16,180 --> 00:03:21,980
Redundancy in the observations; the meaning is, we should have multiple number of
observations.

39
00:03:21,980 --> 00:03:31,040
Then we also talked about the check - we should have something, some observations,
which are
40
00:03:31,040 --> 00:03:37,100
not used in our general plotting work, but
these observations we will use later on.

41
00:03:37,100 --> 00:03:42,489
At the end of this final map, when the map
is produced, when the final results are there

42
00:03:42,489 --> 00:03:46,789
with us, we want to check those results, so
we should have something called a check.

43
00:03:48,240 --> 00:03:51,480
Then, we are talking about plotting of the
details.

44
00:03:52,060 --> 00:03:54,180
We can plot the details by various means.

45
00:03:54,180 --> 00:03:59,290
We saw that, if two points are fixed - we
know this length - so these two points are

46
00:03:59,290 --> 00:04:01,709
fixed, a third point is to be fixed.

47
00:04:02,280 --> 00:04:06,700
What we need to do in order to fix this point: either we measure both the lengths,

48
00:04:07,460 --> 00:04:14,420
or we measure both the angles, or we measure one length and one angle, or maybe one
length

49
00:04:14,430 --> 00:04:19,829
and one angle this way, or maybe - one more possibility - we drop a perpendicular
from

50
00:04:19,829 --> 00:04:26,889
this point on this line . So here, we know
the distance here, along the lines, which

51
00:04:26,889 --> 00:04:32,750
we in our last lecture said X, and this particular distance - perpendicular from
the point - we

52
00:04:32,750 --> 00:04:34,170
said this as offset.

53
00:04:35,300 --> 00:04:40,360
So by measuring this perpendicular and this distance, we can plot this position or
this

54
00:04:40,370 --> 00:04:42,390
point with reference to these two points.

55
00:04:42,960 --> 00:04:44,760
So, we can plot our details that way.

56
00:04:45,300 --> 00:04:50,840
Then, we were talking about very, very important
thing, that is, the plotting accuracy.

57
00:04:50,840 --> 00:04:57,270
Plotting accuracy - in our map, the minimum
point or the minimum thickness of a point

58
00:04:57,270 --> 00:05:02,250
that we can make is, generally, we take it
as 0.25 mm.

59
00:05:02,250 --> 00:05:06,290
What is the meaning of this?

60
00:05:06,290 --> 00:05:11,000
Within that 0.25 mm we cannot see anything,
because it is a single point.

61
00:05:11,000 --> 00:05:15,070
A line - thickness of the line - we cannot
see any detail inside.

62
00:05:15,860 --> 00:05:22,080
Well, if that is the case, the corresponding
distance in the ground - how do you find it?

63
00:05:22,080 --> 00:05:27,150
We find it by multiplying plotting accuracy
by the scale so we know the corresponding

64
00:05:27,150 --> 00:05:28,739
distance there in the ground.

65
00:05:28,740 --> 00:05:34,620
So, the meaning is: because we cannot see
anything within that point in our map, so

66
00:05:34,620 --> 00:05:41,610
we cannot plot any detail which is within
that distance in the ground; the corresponding

67
00:05:41,610 --> 00:05:43,009
plotting accuracy distance.

68
00:05:43,009 --> 00:05:47,660
So we should always keep in mind whenever
you are deciding about your scale: ‘Okay,

69
00:05:47,660 --> 00:05:49,680
my scale of survey is this’.

70
00:05:49,680 --> 00:05:55,520
So you know what kind of details, how accurate
you should observe in the field.

71
00:05:55,520 --> 00:05:58,200
So plotting accuracy is an important thing.

72
00:05:58,200 --> 00:06:03,818
Then finally, we are talking about classification
of survey - how we classify.

73
00:06:03,819 --> 00:06:09,539
There were different classifications: based
on the instrument used, the area - the type

74
00:06:09,539 --> 00:06:15,630
of the area, the functionality - for what
purpose you are using it - so there are different

75
00:06:15,630 --> 00:06:16,630
classifications.

76
00:06:16,630 --> 00:06:19,130
So this is all we have discussed in our last
lecture.

77
00:06:19,900 --> 00:06:26,219
Today, in our this video lecture, we will
be talking about errors in measurement, and
78
00:06:26,220 --> 00:06:30,979
we will be looking into sources of errors,
classification of errors, systematic; random

79
00:06:30,979 --> 00:06:37,270
errors, principle of adjustment, principle
of reversal - maybe, then basic terms, random

80
00:06:37,270 --> 00:06:39,909
error distribution, use of normal distribution
curve and all that.

81
00:06:39,910 --> 00:06:46,729
So, we begin now with our this lecture – so,
we are going to talk about errors in measurement.

82
00:06:47,640 --> 00:06:53,060
We said that in surveying, the measurements are the basic things we need to measure
the

83
00:06:53,060 --> 00:06:53,780
things.

84
00:06:54,699 --> 00:06:59,160
So, we need to also talk about the errors
in measurement.

85
00:06:59,160 --> 00:07:00,910
What is the meaning of this?

86
00:07:00,910 --> 00:07:02,200
Let us start explaining this.

87
00:07:02,200 --> 00:07:07,460
This is a very, very important area and we
must know about this whenever we are doing

88
00:07:07,460 --> 00:07:08,560
the surveying.

89
00:07:08,560 --> 00:07:10,650
Well, let me give you an example.

90
00:07:10,650 --> 00:07:16,690
Here, I have got a scale and in this scale
we have some graduations.

91
00:07:16,690 --> 00:07:22,919
Any length; for example, let us say I want
to measure a length here using this scale.

92
00:07:22,919 --> 00:07:24,520
So, what I will do?

93
00:07:24,520 --> 00:07:30,159
I will put this scale along this length and
I say, ‘Okay, what is this length?’

94
00:07:31,100 --> 00:07:34,180
So, this is a procedure of taking the measurement.

95
00:07:34,189 --> 00:07:37,409
We are taking an observation; we are measuring our length here.

96
00:07:37,940 --> 00:07:44,300
Now, in measuring this length ‘l’, what
do we measure?

97
00:07:45,569 --> 00:07:50,090
Are we measuring the actual value, the true
value of the length?

98
00:07:50,740 --> 00:07:52,189
Now, what is true value?

99
00:07:52,480 --> 00:07:55,699
True value is the value which is actually
there.

100
00:07:55,699 --> 00:07:58,199
Are we able to really measure it with this
scale?

101
00:07:59,560 --> 00:08:07,640
Now, the answer of that is: no, we cannot
measure true value.

102
00:08:07,650 --> 00:08:09,840
Now, why?

103
00:08:09,840 --> 00:08:16,669
One very simple answer of that: this scale
has got a least count, and the least count

104
00:08:16,669 --> 00:08:18,688
of this scale is in millimetres.

105
00:08:21,760 --> 00:08:28,240
We can measure up to 1 millimetre, or we can
estimate within a millimetre; we can approximate

106
00:08:28,240 --> 00:08:29,380
within a millimetre.

107
00:08:30,920 --> 00:08:35,120
Well, when we are approximating, how good
we are in the approximation?

108
00:08:35,120 --> 00:08:38,380
Are we really measuring the true length ‘l’?

109
00:08:38,380 --> 00:08:42,679
No, because we are limited by the least count
of the instrument.

110
00:08:42,679 --> 00:08:51,930
So never, in practice, we can measure the
true value of any length, angle or any geoinformation.

111
00:08:51,930 --> 00:08:54,229
So, what we measure?

112
00:08:54,230 --> 00:09:00,780
We measure the true value - let us say we
are writing it in capital for ‘l’ - plus

113
00:09:01,640 --> 00:09:06,240
some error term, because we are introducing some error here.

114
00:09:06,250 --> 00:09:08,850
So, this is the error term which we are introducing.

115
00:09:10,860 --> 00:09:15,310
I gave you just one example of the error which is because of the least count; there
may be

116
00:09:15,310 --> 00:09:17,030
many more sources of the error.

117
00:09:17,920 --> 00:09:25,780
Well, let us start talking about those - what
are the sources; the sources of error?

118
00:09:25,780 --> 00:09:36,939
Sources of error in measurement - well, let
us talking that.

119
00:09:36,940 --> 00:09:39,900
Number one: are there any

120
00:09:41,720 --> 00:09:51,060
natural sources? Just think about that - let us think about this scale only .This
scale says that this

121
00:09:51,070 --> 00:09:57,370
is one meter, which is written over here - I
can read; from here to here, it says this

122
00:09:57,370 --> 00:09:58,730
is 1 meter.

123
00:09:59,520 --> 00:10:04,840
So, when this scale was manufactured, when this length was given - 1 meter - it was
given

124
00:10:04,840 --> 00:10:06,240
at a certain temperature.

125
00:10:06,840 --> 00:10:13,680
Let us say the temperature at which this scale was given 1 meter was 27 degree
Celsius,

126
00:10:14,560 --> 00:10:20,530
and right now, when I am working in the field, the temperature here is not 27
degree Celsius

127
00:10:20,530 --> 00:10:26,150
- rather, more than that, 42 degree Celsius.

128
00:10:27,700 --> 00:10:29,900
Now what happens because of this?

129
00:10:30,120 --> 00:10:35,680
If the material of this scale is such that
it expands - all the materials will -

130
00:10:36,320 --> 00:10:38,880
it should expand with the temperature.

131
00:10:39,880 --> 00:10:44,800
Still, I see it is written 1 meter here, isn’t
it?

132
00:10:44,810 --> 00:10:50,510
But because of the temperature, the actual
length of this scale is not 1 meter, rather,

133
00:10:50,510 --> 00:10:52,200
slightly more than 1 meter.

134
00:10:52,200 --> 00:11:00,950
Now, if I measure any length with this scale,
which has expanded, our length measurement

135
00:11:00,950 --> 00:11:04,950
will not be accurate, rather, we are introducing
some error.

136
00:11:04,950 --> 00:11:09,980
So this is one example of natural source of
error.

137
00:11:09,980 --> 00:11:12,830
There could be many more - I can give you
one more example.

138
00:11:12,830 --> 00:11:17,520
For example, let us say, many times you must
have seen, as I am drawing here in the diagram

139
00:11:19,040 --> 00:11:24,160
well, what I do,
I suspend a heavy plumb-bob.

140
00:11:24,160 --> 00:11:36,829
What this plumb-bob is: here is the thread
or twine and here is a heavy bob, a very heavy

141
00:11:36,830 --> 00:11:37,830
material.
142
00:11:37,830 --> 00:11:43,880
We make use of this in order to find the direction
of the gravity; obviously, because of this

143
00:11:43,880 --> 00:11:47,620
weight, it will align in the direction of
the gravity.

144
00:11:47,620 --> 00:11:51,150
So, whenever we want to measure the direction
of the gravity, we can make use of this simple

145
00:11:51,150 --> 00:11:52,150
device.

146
00:11:52,150 --> 00:11:57,230
We want to ensure whether a particular wall
is vertical or not - you must have seen many

147
00:11:57,230 --> 00:12:02,110
masons using this plumb-bob to ensure the
verticality.

148
00:12:02,780 --> 00:12:05,699
Now, how measure is coming into the picture?

149
00:12:07,120 --> 00:12:13,100
If there is some wind is blowing - it is not
appreciable, but the wind is blowing there

150
00:12:13,100 --> 00:12:15,910
- because of the wind blowing, what will happen?

151
00:12:15,910 --> 00:12:24,810
A constant force will be acting on the plumb-bob
and the plumb-bob will be now slightly displaced.

152
00:12:26,230 --> 00:12:32,990
I am drawing slightly exaggerated one; it
will not be this big, but it will be slightly

153
00:12:32,990 --> 00:12:35,790
swayed from its vertical position.

154
00:12:36,280 --> 00:12:37,339
So, what we are going to do?

155
00:12:37,350 --> 00:12:41,800
We are not able to measure the exact perpendicular
or the vertical direction; the direction of

156
00:12:41,800 --> 00:12:44,991
the gravity; rather, we are measuring something
else.

157
00:12:44,991 --> 00:12:48,670
So again, the nature is playing a role here.

158
00:12:48,670 --> 00:12:53,120
So, whenever we take the measurements, we
have sources of the errors which are because

159
00:12:53,120 --> 00:12:54,200
of the nature.

160
00:12:54,200 --> 00:13:04,810
Then, the second one: the second source of
the error is because of the instrument.

161
00:13:04,810 --> 00:13:10,810
Well, how - how come?

162
00:13:11,370 --> 00:13:15,760
I gave you one example that in this scale,
if the least count - there is a limitation

163
00:13:15,760 --> 00:13:16,760
of the least count.

164
00:13:17,420 --> 00:13:23,740
Let us say the least count in this case is
1 centimetre -in one case.

165
00:13:23,750 --> 00:13:27,370
In other case, the least count is 1 millimetre.

166
00:13:29,380 --> 00:13:32,380
Now can you start thinking the limitations?

167
00:13:32,390 --> 00:13:39,470
In this case, we can measure more precisely, more closely, while in this case, our
measurement

168
00:13:39,470 --> 00:13:41,970
would not be that close.

169
00:13:41,970 --> 00:13:43,650
So, what is happening?

170
00:13:43,650 --> 00:13:47,370
Because of the instrument itself, there are
some errors being introduced.

171
00:13:47,700 --> 00:13:55,820
I will give you one more example: well, this
says ‘1 metre’ - is it so?

172
00:13:55,820 --> 00:13:57,700
The question is, is it really 1 metre?

173
00:13:57,700 --> 00:14:02,990
Because this has been manufactured by someone;
someone has manufactured an instrument, which

174
00:14:02,990 --> 00:14:09,610
may be the scale - the simple one like this
- or maybe a sophisticated instrument.

175
00:14:09,610 --> 00:14:15,270
And this instrument says this is 1 metre,
but is it really so?

176
00:14:15,270 --> 00:14:16,270
That is the big question.

177
00:14:16,270 --> 00:14:22,930
So, if it is not really so, it says it is
1 metre, but actually, if you measure it with

178
00:14:22,930 --> 00:14:27,180
some standard, you will find it is 1.01 metre.

179
00:14:27,180 --> 00:14:35,489
So we have some error included in our instrument,
and whatever the length, we will measure with

180
00:14:35,490 --> 00:14:36,690
this ruler.

181
00:14:36,690 --> 00:14:38,480
So what we are doing?

182
00:14:38,480 --> 00:14:45,940
The actual length of this is 1.01 metre, but
we are considering this to be 1 metre.

183
00:14:48,160 --> 00:14:54,110
The meaning is: if, here in the ground - let
us say this length - if this length is 1.01

184
00:14:54,110 --> 00:15:02,450
metre, and if you are measuring it with this
scale which has actual length of 1.01 metre,

185
00:15:02,450 --> 00:15:11,600
then this length, which is actually 1.01 metre will be shown as 1 metre, isn’t it?

186
00:15:11,600 --> 00:15:15,620
Because our scale is showing 1.01 metre as
1 metre.

187
00:15:15,620 --> 00:15:19,960
So this is one source of error which is coming because of the instrument.

188
00:15:21,460 --> 00:15:27,220
Well, the third one - third one you can guess
very well.

189
00:15:27,230 --> 00:15:28,230
What is the third source of the error?

190
00:15:28,230 --> 00:15:33,100
It is because of the person, or ‘personal
error’.

191
00:15:33,100 --> 00:15:38,760
The person who is working there - well, I
do not want - I am a bit tired; I am not happy

192
00:15:38,760 --> 00:15:40,860
today, and I am taking the measurement.

193
00:15:41,680 --> 00:15:47,120
The weather is not very good; it is very hot,
so I am not very, you know, happy about working.

194
00:15:47,120 --> 00:15:48,120
So what do I do?

195
00:15:48,120 --> 00:15:50,590
I start committing a mistake when I take the
measurement.

196
00:15:51,240 --> 00:15:57,400
Also, you know when we take the observations over a scale, where our eyes should be
- our

197
00:15:57,430 --> 00:16:03,810
eyes should be exactly over the graduation
or exactly over the mark for which you want

198
00:16:03,810 --> 00:16:08,319
to measure the length, and then we see the
graduation corresponding to that.

199
00:16:08,320 --> 00:16:13,090
Well, one source of error could be if I displace
my head here or here - the parallax.

200
00:16:14,000 --> 00:16:15,140
So what is happening?

201
00:16:15,140 --> 00:16:19,400
Now, the person who is working with the instrument
is introducing some errors.

202
00:16:19,410 --> 00:16:22,290
So, there are errors because of the person
also.

203
00:16:23,520 --> 00:16:29,149
Well, having seen all these types of the errors, what we will do now, we will go
into another

204
00:16:29,149 --> 00:16:33,949
classification which we say, ‘classification
of errors’.

205
00:16:35,000 --> 00:16:40,960
So far we are talking about the sources of
the error; now we will see classification

206
00:16:40,970 --> 00:16:42,360
of errors.

207
00:16:42,360 --> 00:16:45,140
What are their types?

208
00:16:45,140 --> 00:16:55,220
What I will do, I will start with one example,
let us say, the number one case, because it

209
00:16:55,220 --> 00:17:02,720
is the number one type of error we say to
be blunder or mistake.

210
00:17:07,119 --> 00:17:14,760
Well, the example in this case is: let us
say, this is the target - many of you must

211
00:17:14,760 --> 00:17:17,619
have done the firing in your NCC classes or
somewhere.

212
00:17:18,440 --> 00:17:25,460
So, when you are firing, what you have try
to do, you have a little bulls eye here.

213
00:17:25,470 --> 00:17:31,110
Your aim is to fire at the point here in the
bulls eye.

214
00:17:31,110 --> 00:17:38,918
Well, if you are careless - careless means,
you are not even targeting the bulls eye - what

215
00:17:38,919 --> 00:17:45,500
you might end up, you might end up firing
your bullet somewhere here or outside or very

216
00:17:45,500 --> 00:17:48,470
much outside, isn’t it?

217
00:17:48,470 --> 00:17:49,470
So what you are doing?

218
00:17:49,470 --> 00:17:56,750
Actually, the person who is taking the observations
is committing some mistakes; there are some

219
00:17:56,750 --> 00:17:57,750
blunders.

220
00:17:57,750 --> 00:17:59,580
So that kind of thing, we say the ‘blunder’.

221
00:17:59,580 --> 00:18:02,809
Now, how it can happen if I am measuring this
length?

222
00:18:02,809 --> 00:18:09,500
Well, I measure this length, I find this to
be 30 - actually, the length there in the

223
00:18:09,500 --> 00:18:13,110
ground is 30.7 centimetre.

224
00:18:13,110 --> 00:18:23,689
Well, I record it, I say, ‘Well, I have
observed it 30.7’ Now, when I write this

225
00:18:23,690 --> 00:18:33,430
length in my notebook, length measurement number one - I write it as, not 30; 13.7.

226
00:18:34,879 --> 00:18:38,969
It might happen; it does happen, because whenever you are working in the field, you
know, if

227
00:18:38,970 --> 00:18:44,020
the conditions to work in the field are not
very, very good - which is normally so, because

228
00:18:44,020 --> 00:18:49,360
you do the survey out there in the ground;
the conditions are not very favourable always.

229
00:18:50,420 --> 00:18:55,940
So, your mind is a bit tired, so you end up
writing a length which is 30.7, which you

230
00:18:55,940 --> 00:18:58,360
also measured, as 13.7.
231
00:18:59,279 --> 00:19:06,080
So what you are doing actually, you are introducing an error, which we will say
blunder or mistake.

232
00:19:06,080 --> 00:19:13,000
Well, the second classification of the error:
the second one of this, we say ‘systematic error’.

233
00:19:14,340 --> 00:19:24,139
Systematic error -
now what happens in this case?

234
00:19:25,420 --> 00:19:27,480
The same example of the target here.

235
00:19:29,100 --> 00:19:35,340
Well, the bulls eye; you are trying to fire
here.

236
00:19:35,340 --> 00:19:41,570
Let us say you are not aware about the gravity
- you are a very good fireman; very good fireman,

237
00:19:41,570 --> 00:19:43,250
but you do not know that the gravity exists.

238
00:19:43,250 --> 00:19:50,250
So, what you are doing, you are trying to
target at the centre, but because of the gravity,

239
00:19:50,250 --> 00:19:55,080
all your bullets are coming here - down.

240
00:19:55,080 --> 00:20:00,610
Maybe a little one will go there, but most
of the bullets will be coming here.

241
00:20:00,610 --> 00:20:04,059
Is it or not?

242
00:20:04,059 --> 00:20:05,059
What is happening now?

243
00:20:05,059 --> 00:20:11,610
The gravity is playing its role; so a systematic
force in the nature is bringing your bullets
244
00:20:11,610 --> 00:20:17,469
down, and your bullets are firing here, not
at the target.

245
00:20:17,470 --> 00:20:23,210
So, this is a systematic force which is working;
so the error is, we say, systematic error.

246
00:20:23,210 --> 00:20:30,799
Similarly, for example, let us say we are
measuring with this scale, and we know this

247
00:20:30,799 --> 00:20:38,830
scale was made at a temperature of 27 degree
Celsius, while the temperature now is 41 degree

248
00:20:38,830 --> 00:20:41,009
Celsius while I am using it.

249
00:20:41,009 --> 00:20:45,260
Obviously, this scale has some expansion in
its length.

250
00:20:46,340 --> 00:20:48,820
So, what is happening in all the measurements?

251
00:20:48,830 --> 00:20:54,389
Let us say I use this scale to measure this
length, and I keep this scale once, twice,

252
00:20:54,390 --> 00:20:57,549
thrice, four times, and so on.

253
00:20:57,549 --> 00:20:58,980
So, what I am doing?

254
00:20:58,980 --> 00:21:05,700
I am introducing some error always, everywhere.

255
00:21:05,700 --> 00:21:12,299
So this kind of error, for which - you saw
it in the case of the gravity, in the case

256
00:21:12,299 --> 00:21:21,839
of the temperature - for which we can write
the mathematical model or, we can say,

257
00:21:23,020 --> 00:21:30,980
those errors which follow some physical law, as in the case of temperature.

258
00:21:31,769 --> 00:21:36,529
There are many, many cases where, you know, you can write a mathematical model for
the

259
00:21:36,529 --> 00:21:39,889
error, because it follows some physical law.

260
00:21:39,889 --> 00:21:41,949
So you can eliminate that error.

261
00:21:41,950 --> 00:21:44,470
Those errors we say ‘systematic error’.

262
00:21:46,500 --> 00:21:55,360
Well, then number three: this is very interesting error, now, and we say this error
to be random

263
00:21:59,600 --> 00:22:01,279
- random error in our observation.

264
00:22:03,280 --> 00:22:08,300
For example, let us say - again, I am going
back to our target.

265
00:22:08,300 --> 00:22:14,120
You are trying to fire at the bulls eye, which
is this point.

266
00:22:14,120 --> 00:22:17,750
Your aim is to fire at the centre here.

267
00:22:17,750 --> 00:22:24,080
Well, what we do, we ensure there is no blunder;
no mistake.

268
00:22:24,080 --> 00:22:29,549
Your firing instructor is very, very strict;
he is ensuring that there is no mistake or

269
00:22:29,549 --> 00:22:30,549
no blunder.
270
00:22:30,549 --> 00:22:35,499
You are not firing here and there; at the
same time you are also aware that the gravity

271
00:22:35,499 --> 00:22:36,499
is there.

272
00:22:36,499 --> 00:22:37,499
So, what you are doing?

273
00:22:37,499 --> 00:22:42,240
You are taking some precautions and you are
raising your gun a little bit, so that instead

274
00:22:42,240 --> 00:22:50,139
of firing at this point, you fire somewhere
here, so that you are taking care of the gravity

275
00:22:50,139 --> 00:22:51,139
also.

276
00:22:51,139 --> 00:22:53,779
Now, if this is the case, what will happen?

277
00:22:53,779 --> 00:23:01,710
All your measurements, all your bullets, will
now fire like this.

278
00:23:01,710 --> 00:23:09,029
Maybe some are farther also, but most of them
will be nearer.

279
00:23:09,029 --> 00:23:14,769
Now this is very interesting observation;
very, very interesting observation: what we

280
00:23:14,769 --> 00:23:22,789
see here, these bullets are spread throughout
my bulls eye, while more number of them are

281
00:23:22,789 --> 00:23:27,949
concentrated in the bulls eye and less are
in the outer periphery.

282
00:23:29,340 --> 00:23:35,699
Now, one thing: why we could not fire in the
bulls eye exactly?

283
00:23:35,700 --> 00:23:36,700
Can we?

284
00:23:36,700 --> 00:23:40,169
No, still there are many sources of errors
which we do not know.

285
00:23:40,169 --> 00:23:45,850
You are breathing, some little turbulence
in the air, some little problem with the gun,

286
00:23:45,850 --> 00:23:51,519
some little problem with the bullet itself
- it does not go in the straight path.

287
00:23:51,519 --> 00:23:55,730
So, there are many sources of errors which
you cannot account for, and these are the

288
00:23:55,730 --> 00:24:00,840
sources of errors which may go in positive
direction or in negative direction; we do

289
00:24:00,840 --> 00:24:02,350
not know how they varying.

290
00:24:02,350 --> 00:24:07,629
So, that kind of errors; those sources of
errors, they contribute the error which is

291
00:24:07,629 --> 00:24:11,820
called random error.

292
00:24:11,820 --> 00:24:12,820
What we observe here?

293
00:24:12,820 --> 00:24:23,119
The random error, their distribution is - number one point about them - they are
equally distributed

294
00:24:23,119 --> 00:24:27,879
in positive and negative directions.

295
00:24:27,879 --> 00:24:31,379
What is the meaning?

296
00:24:31,379 --> 00:24:36,860
Over here, there are random errors in all
the directions.

297
00:24:36,860 --> 00:24:40,000
It is not that they are in only one direction;
they are in all the directions.

298
00:24:40,000 --> 00:24:45,690
So, their probability of occurrence in positive
direction and negative direction is same.

299
00:24:45,690 --> 00:24:48,870
Well, the example here will be the ruler;
the scale.

300
00:24:48,870 --> 00:24:54,498
While I am measuring this length, I am trying
to keep my scale exactly over the point here

301
00:24:54,499 --> 00:24:55,499
and here.

302
00:24:55,499 --> 00:24:58,400
What is the guarantee that I am keeping exactly
over it?

303
00:24:58,400 --> 00:25:02,230
Again, there is some subjectivity involved;
the human judgement is involved.

304
00:25:02,230 --> 00:25:08,290
I may be keeping it slightly here or slightly
here, so that introduces the random error.

305
00:25:08,290 --> 00:25:11,178
So, what will happen?

306
00:25:11,179 --> 00:25:17,179
The error amount - now number two, point number
two about it - what we see here: random error

307
00:25:17,179 --> 00:25:22,169
which are large in magnitude - for example,
this - occur very less.

308
00:25:22,169 --> 00:25:30,590
There is less possibility, of the random errors
which are large in amount, of occurring, but

309
00:25:30,590 --> 00:25:35,718
the random error which is small, for example,
here - here, we have a huge concentration

310
00:25:35,719 --> 00:25:39,570
- the chances of their occurring are more.

311
00:25:39,570 --> 00:25:59,619
So, we can say: large random errors, they
occur less, while small random errors, they occur

312
00:26:00,720 --> 00:26:02,060
more.

313
00:26:03,040 --> 00:26:09,860
We will see in a moment some further things about this random error, but this is
one error,

314
00:26:09,869 --> 00:26:15,549
you know, if you eliminate from your observation - let us say, your observation is
‘l’.

315
00:26:15,880 --> 00:26:20,160
For a length which you needed, you measure it as ‘l’.

316
00:26:20,169 --> 00:26:24,119
You eliminate that this is not blunder or
mistake - fine.

317
00:26:24,440 --> 00:26:32,000
Then, you eliminate some systematic errors - systematic error ‘SE’ - from your
observation.

318
00:26:32,730 --> 00:26:35,200
You know the model how this error has occurred.

319
00:26:35,200 --> 00:26:37,879
So, you eliminate this systematic error also.

320
00:26:37,879 --> 00:26:41,998
Now, in your observation, what about the outcome
of this?

321
00:26:41,999 --> 00:26:45,169
For example, let us say the outcome is l’.

322
00:26:45,169 --> 00:26:55,899
Still, this l’ is not equal to true length,
because your l’ has still some errors, and

323
00:26:55,899 --> 00:26:57,799
those are the errors which we say random errors.

324
00:26:57,799 --> 00:27:02,899
So, we can say the random errors are those
errors which we made in observation even after

325
00:27:02,899 --> 00:27:06,498
eliminating systematic errors from the observation.

326
00:27:07,110 --> 00:27:12,550
Now, what we will do, we will discuss two
more terms here. One term is called

327
00:27:15,600 --> 00:27:16,719
‘precision’.

328
00:27:17,400 --> 00:27:18,440
What is the meaning of that?

329
00:27:18,450 --> 00:27:22,269
Again, I am going back to our example of target.

330
00:27:22,269 --> 00:27:29,929
That is the bulls eye, we are trying to target
the bulls eye somewhere here,

331
00:27:30,529 --> 00:27:34,629
but you are not able to do that. Precision means,

332
00:27:36,040 --> 00:27:41,940
let us say, when we are trying
to target here, there are some systematic

333
00:27:41,950 --> 00:27:48,731
errors also working, and because of the systematic errors, most of your
observations, or we can

334
00:27:48,731 --> 00:27:53,389
say, the bullets, are clubbed in this area.

335
00:27:53,389 --> 00:27:58,928
It might happen; let us say some force is
working upward now - not like the gravity

336
00:27:58,929 --> 00:28:01,849
- and most of the observations are now in
this area.

337
00:28:03,840 --> 00:28:09,139
Well, this is one case - this is the - a person
A fired these bullets.

338
00:28:09,149 --> 00:28:18,089
Person B also fired these bullets, and for
person B, what he gets, he gets all the bullets,

339
00:28:18,400 --> 00:28:19,880
let us say, like this.

340
00:28:22,460 --> 00:28:26,040
Well, what would we like to say about person A and person B?

341
00:28:26,049 --> 00:28:30,429
Who is better firer; who is better fireman;
who has got better aim?

342
00:28:31,580 --> 00:28:36,699
Despite there is some systematic error - which is, maybe, your barrel is slightly
tilted;

343
00:28:37,020 --> 00:28:44,220
though you are targeting the bulls eye, all
your bullets are being fired here, slightly

344
00:28:44,230 --> 00:28:52,570
swayed, but this person has all his bullets
very close to each other; while this person

345
00:28:52,570 --> 00:28:55,669
has his bullets, you know, widespread.

346
00:28:56,220 --> 00:29:07,580
So, definitely, we can say person B is more
precise in his firing than person A. This

347
00:29:07,580 --> 00:29:21,059
explains the term precision: we mean by precision the closeness of observation; how
close your

348
00:29:21,059 --> 00:29:22,299
observations are.

349
00:29:22,600 --> 00:29:29,659
Another example: well, in this scale - the
least count of this scale is 1 centimetre.

350
00:29:30,809 --> 00:29:37,570
You are trying to measure up to 1 centimetre and then you are approximating within
1 centimetre

351
00:29:37,570 --> 00:29:38,570
- that is case number one.

352
00:29:38,920 --> 00:29:46,900
In case number two, the least count is 1 millimetre . So, now you are confident to
measure up

353
00:29:46,909 --> 00:29:51,909
to the millimetre, and you are estimating
only within 1 millimetre.

354
00:29:51,909 --> 00:29:55,390
Obviously, in this case, the precision is
more.

355
00:30:00,460 --> 00:30:03,280
Then, compare to the other case.

356
00:30:06,400 --> 00:30:11,340
Another term which is
related with the precision is accuracy.

357
00:30:15,640 --> 00:30:17,560
Well, what is accuracy?

358
00:30:18,780 --> 00:30:29,940
Let us go back to again our target - the bulls eye here; you are trying to fire at
the centre.
359
00:30:32,060 --> 00:30:43,060
Now, in one case, you fire here - let us say
the fireman A, he fires here.

360
00:30:43,590 --> 00:30:50,250
In the second case, there is another fireman, and he fires here.

361
00:30:54,780 --> 00:30:59,000
Now, out of A and B, who is more accurate?

362
00:30:59,779 --> 00:31:00,779
What is the meaning of accuracy?

363
00:31:00,779 --> 00:31:06,570
Obviously, you will say B is more accurate
because he is firing nearer to the target,

364
00:31:06,570 --> 00:31:09,710
nearer to the bulls eye, while A is not that
accurate.

365
00:31:10,260 --> 00:31:25,200
Now, in terms of precision, both A and B are same; the precision spread is same,
but A

366
00:31:25,200 --> 00:31:26,840
is not accurate.

367
00:31:27,320 --> 00:31:28,720
So what is the meaning of accuracy?

368
00:31:28,720 --> 00:31:37,600
The meaning of accuracy is, closeness of your observation to the true value; how
close your

369
00:31:37,600 --> 00:31:40,000
observations are to the true value.

370
00:31:40,440 --> 00:31:46,920
But, there may be a very interesting case;
in that case, in this bulls eye -

371
00:31:48,440 --> 00:31:50,000
now, there are two persons.
372
00:31:52,420 --> 00:32:03,680
One person has got his bullets here, the other person has got his bullets here.

373
00:32:06,009 --> 00:32:13,909
You should take the mean of these - the mean is here somewhere - that is for
observer A,

374
00:32:13,909 --> 00:32:19,429
and for the second person B, the mean will
be somewhere here.

375
00:32:20,040 --> 00:32:27,639
Even if the precision is more in this case,
it is not accurate.

376
00:32:27,649 --> 00:32:35,469
Here the precision is less - your observations
are far apart - but your accuracy is more.

377
00:32:35,470 --> 00:32:40,740
But if you eliminate the systematic error
from your observations, if the systematic

378
00:32:40,740 --> 00:32:43,820
error is eliminated, then what will happen?

379
00:32:43,820 --> 00:32:48,918
All these bullets, they will come to the target
or the bulls eye.

380
00:32:48,919 --> 00:32:56,769
So, if the systematic errors are eliminated
in that case, the precision of the observations

381
00:32:56,769 --> 00:32:58,440
becomes an indicator of the accuracy.

382
00:32:58,440 --> 00:33:04,700
Now, what we will do, we will look at one
more important principle here that is called

383
00:33:04,700 --> 00:33:11,710
‘principle of adjustment’.

384
00:33:11,710 --> 00:33:20,859
Now, we know by this time whatever we are
measuring - for example, let us say there

385
00:33:20,859 --> 00:33:31,168
is a triangle, and for this triangle we are
measuring three angles: theta 1, theta 2 and

386
00:33:31,169 --> 00:33:32,799
theta 3.

387
00:33:32,799 --> 00:33:41,520
There are three points there in the ground
- one point, another point, another point,

388
00:33:41,520 --> 00:33:48,590
and using some equipment, some instrument,
I stand at point 1 - number 1 - I measure

389
00:33:48,590 --> 00:33:49,590
the theta 1.

390
00:33:49,590 --> 00:33:52,629
Similarly, I measure theta 2, theta 3; all
the angles.

391
00:33:52,629 --> 00:34:01,279
Now, we know that - this theta 2, for example
- it is not really the true value.

392
00:34:01,279 --> 00:34:05,799
The true value, in this case, is ‘theta
2 hat’, let us say.

393
00:34:05,799 --> 00:34:11,630
Let us say this is the true value, but theta
2 is not the true value - why?

394
00:34:11,630 --> 00:34:15,890
Because theta 2 has got some errors in it.

395
00:34:15,889 --> 00:34:18,460
We have discussed this thing that we cannot
measure the true value.

396
00:34:19,800 --> 00:34:21,920
So, all our theta 2 has got the error.

397
00:34:22,840 --> 00:34:29,180
Now, if that is the case - well, because this
is the triangle, I can write now as ‘sigma

398
00:34:29,190 --> 00:34:40,101
theta head’ the true angle - the true angles
between these three points should come to

399
00:34:40,101 --> 00:34:41,101
180.

400
00:34:41,101 --> 00:34:45,910
Obviously, the true angles, if we can measure
true angles - the ‘theta hat’ - they should

401
00:34:45,909 --> 00:34:54,040
come to 180 degree, but if you are writing
‘sigma theta’, the observations - will

402
00:34:54,040 --> 00:34:59,390
it be equal to 180?

403
00:34:59,390 --> 00:35:04,420
No, because all our observations have got
some error also.

404
00:35:04,420 --> 00:35:10,280
Well, there in the field there are three points
for which we have measured these angles.

405
00:35:10,280 --> 00:35:13,700
The angles have got the error, but the field
does not.

406
00:35:13,700 --> 00:35:17,919
The angles which I have measured have got
the error, so what I need to do, I need to

407
00:35:17,920 --> 00:35:23,190
correct my angles so that they satisfy the
condition of the field.

408
00:35:23,190 --> 00:35:27,340
The condition of this field is that these
three angles which I have observed should

409
00:35:27,340 --> 00:35:28,890
come to 180.

410
00:35:28,890 --> 00:35:30,810
So, what I need to do?

411
00:35:30,810 --> 00:35:39,610
I need to apply some corrections to the angles,
and we do it by the procedure of adjustment.

412
00:35:39,610 --> 00:35:44,320
We will see in detail about this adjustment
later on.

413
00:35:44,320 --> 00:35:49,040
So, now we will talk something more about
the errors; some more terms.

414
00:35:53,060 --> 00:35:59,660
Well, the very first term I would like to
see is true error.

415
00:36:00,680 --> 00:36:08,180
We have already seen the term ‘true value’
- we cannot measure the true value; there

416
00:36:08,190 --> 00:36:12,200
are limitations - so many errors come in to
the observation, so we cannot measure the

417
00:36:12,200 --> 00:36:13,200
true value.

418
00:36:14,320 --> 00:36:16,780
Well, so, do we know the true error then?

419
00:36:18,040 --> 00:36:23,600
If I know the true value, then only, I will
know the true error, because the true error

420
00:36:23,609 --> 00:36:28,288
- for example, let us say there is a length;
its true value is ‘L’.

421
00:36:29,310 --> 00:36:31,120
What I observe is ‘l’.

422
00:36:31,700 --> 00:36:40,140
So, the error - if I write error as ‘e’,
and the true error will be ‘L minus l’.

423
00:36:41,420 --> 00:36:47,680
But I do not know L - if I do not know L,
just by using the observation, I cannot find

424
00:36:47,690 --> 00:36:49,190
the true error.

425
00:36:50,600 --> 00:36:51,960
Well, what we can find?

426
00:36:51,970 --> 00:37:01,990
We can find - I am going to give you another
term that is called ‘relative error’.

427
00:37:06,640 --> 00:37:11,540
In some cases, for example, here in the case
of the scale, I know what kind of error is

428
00:37:11,550 --> 00:37:15,860
being introduced because of the least count;
I know that thing.

429
00:37:15,860 --> 00:37:23,680
Let us say, that error which is being introduced because of the least count is
‘delta l’.

430
00:37:24,520 --> 00:37:36,600
Now, if I observe a length of line which is
here - l1 - I can find delta l by l1, and

431
00:37:36,610 --> 00:37:38,250
this we say as the relative error.

432
00:37:38,250 --> 00:37:40,550
Now, why do we need it?

433
00:37:40,550 --> 00:37:47,280
We need it because, using the same scale,
I can also measure a length which is l2 - in

434
00:37:47,280 --> 00:37:51,940
that case, the relative error will become
delta l by l2.
435
00:37:52,540 --> 00:37:56,700
So, we can compare our observations; compare
the error.

436
00:37:56,700 --> 00:38:00,160
So this is how this relative error is used.

437
00:38:00,160 --> 00:38:06,100
Many times, this is also called ‘percentage
error’ , where you simply multiply this

438
00:38:06,100 --> 00:38:10,279
by 100, and that becomes the percentage error.

439
00:38:10,280 --> 00:38:12,230
Okay.

440
00:38:12,230 --> 00:38:27,870
Now, we will talk about one very important
concept that is called ‘Most Probable Value’

441
00:38:30,060 --> 00:38:32,360
or MPV.

442
00:38:33,000 --> 00:38:37,580
We do not know true value; yes, we do not
know that.

443
00:38:37,590 --> 00:38:38,750
What is most probable value?

444
00:38:40,240 --> 00:38:45,200
It is a kind of replacement for the true value
which we use mostly in our cases.

445
00:38:45,210 --> 00:39:00,050
The most probable value is defined as one
which has maximum likelihood to be nearest

446
00:39:00,050 --> 00:39:08,150
to the true value . Well, let us explain this.

447
00:39:08,150 --> 00:39:15,370
A value which has maximum probability of being
nearest to the true value - you know we cannot
448
00:39:15,370 --> 00:39:17,049
get the true value by any means.

449
00:39:17,050 --> 00:39:20,270
But, what we can do, we can go nearer to the
true value.

450
00:39:20,270 --> 00:39:21,690
I will give one example here.

451
00:39:21,880 --> 00:39:29,660
Let us say there is a length L - L is the true value which we cannot measure. What
we are doing?

452
00:39:29,660 --> 00:39:38,020
We are observing this: in one case, l1, l2, l3 - we have taken three observations .

453
00:39:38,220 --> 00:39:47,080
In another case, we are taking from l1 to l100 - hundred observations of the same
length - more redundancy

454
00:39:47,080 --> 00:39:53,020
here in this case; less redundancy here. Now, by taking the mean - arithmetic mean
- of these,

455
00:39:53,480 --> 00:40:00,740
I will get an ‘l bar’ of case number
1, I will get - there also - an ‘l bar’

456
00:40:00,740 --> 00:40:07,520
of case number 2; there are two cases. So, obviously, just by common sense right
now,

457
00:40:07,520 --> 00:40:11,100
you can say that the probability of l2

458
00:40:12,200 --> 00:40:15,020
being nearer to the true value is more.

459
00:40:16,240 --> 00:40:24,379
So, we can say this is our most probable value or rather, the other way round: out
of all these 100

460
00:40:24,380 --> 00:40:25,220
measurements,
461
00:40:27,580 --> 00:40:33,819
which measurement is most probable?
Is it l1, l2, l3, or l100 ? We do not know about it,

462
00:40:34,560 --> 00:40:39,560
but once we have the observation
which is the mean, we can say this l2 has

463
00:40:39,560 --> 00:40:47,160
the maximum probability of being nearer to the true value. So this l2 is our most
probable value.

464
00:40:48,280 --> 00:40:54,440
What we are saying in this case, we are saying that the most probable value is the
arithmetic mean

465
00:40:59,140 --> 00:41:06,460
this is how we take it mostly. If I write the most probable value for some
observation as ‘X bar’,

466
00:41:06,960 --> 00:41:14,460
so the meaning of that is, it is sigma xi by n, where i varies from 1 to n.

467
00:41:15,520 --> 00:41:22,140
For anything - any quantity, any variable - there are n number of observations -
x1, x2, x3 -

468
00:41:22,140 --> 00:41:24,900
which we are writing as xi here, and that is the arithmetic mean.

469
00:41:25,600 --> 00:41:33,000
And we are saying that the arithmetic mean is the most probable value.

470
00:41:35,080 --> 00:41:38,299
Is it, really so - can we prove it? Now, to prove it,

471
00:41:40,660 --> 00:41:43,020
we will do a little exercise,

472
00:41:45,320 --> 00:41:54,140
and I will define one more term before that -the term is ‘residual’.

473
00:41:57,160 --> 00:42:05,819
I write the residual as ‘r’, and this
is written as ‘X bar minus x’
474
00:42:07,540 --> 00:42:14,320
for any observation I. So, what is the residual? Residual is

475
00:42:15,220 --> 00:42:17,379
the mean of the observation or - sorry,

476
00:42:18,140 --> 00:42:26,040
X bar we are taking as most probable value; let the X bar be the most probable
value

477
00:42:27,760 --> 00:42:34,260
which has maximum chance of being nearest to the true value. If that is so, in the
case of the X bar,

478
00:42:34,600 --> 00:42:39,080
we define the residual as the most
probable value minus the observation.

479
00:42:39,940 --> 00:42:45,420
So, for all the observations - i is equal to 1
to n - we can find their corresponding residual.

480
00:42:46,340 --> 00:42:52,420
So this is called the residual. Now, we will
make use of this in order to prove that

481
00:42:52,620 --> 00:42:57,100
the most probable value is nothing but arithmetic mean of the observation.

482
00:42:57,520 --> 00:43:02,800
So, we had written that our ri or the residual was

483
00:43:03,140 --> 00:43:11,279
X bar minus xi. What I do, I sum it up for all the observations: let us say

484
00:43:11,920 --> 00:43:19,780
sigma ri for i is equal to 1 to n will be sigma

485
00:43:20,120 --> 00:43:26,740
X bar minus sigma xi; i equal to 1 to n .

486
00:43:29,460 --> 00:43:37,000
Well, now look at this part - this is very important. The residual -

487
00:43:37,780 --> 00:43:39,920
what will be the sign of residual?

488
00:43:43,960 --> 00:43:49,640
We are talking here, now, only in the case when all our observations,

489
00:43:51,140 --> 00:43:53,759
our variable, has

490
00:43:57,420 --> 00:44:04,380
only random error. Let us be clear about it

491
00:44:04,680 --> 00:44:10,000
whatever we are talking here, our variable, our observation, has only the random
error

492
00:44:10,000 --> 00:44:15,500
in it; we have already eliminated the blunder, we have taken care of the systematic
error

493
00:44:15,500 --> 00:44:19,960
from the observations, and our observation have got only the random error.

494
00:44:20,440 --> 00:44:26,680
We are talking in that case, and only in that case we are saying that the most
probable value

495
00:44:26,940 --> 00:44:29,420
will be the arithmetic mean. And this is what we are going to prove.

496
00:44:30,360 --> 00:44:36,580
And only in that case, we are defining the residual, which is the most probable
value minus

497
00:44:36,780 --> 00:44:42,700
the observation. Well once that is clear, we are looking into this part.

498
00:44:43,960 --> 00:44:50,240
What this part will be, now? For a large n -

499
00:44:52,240 --> 00:44:59,459
many observations - you know what is the
sign of r, the residual? The r may be in positive

500
00:44:59,460 --> 00:45:05,780
direction as well as in negative direction,
which we saw. The random errors have got equal

501
00:45:05,780 --> 00:45:09,520
probability of going into the positive direction as well as in the negative
direction - this

502
00:45:09,520 --> 00:45:15,740
is what we have seen already. So, the sign
of r will be positive and negative with equal

503
00:45:15,740 --> 00:45:23,100
probability. If that is so, for large n, this
particular part will be equal to 0.

504
00:45:25,020 --> 00:45:37,259
If that is so, what we can write now? Sigma ri is sigma x bar minus sigma xi.

505
00:45:37,640 --> 00:45:47,759
What we can write if this becomes 0? You can write
it as: sigma X bar is sigma xi. What is sigma

506
00:45:47,760 --> 00:45:57,180
X bar? Sigma X bar is n times the X bar is
being summed up. So, that is equal to sigma

507
00:45:57,180 --> 00:46:06,210
xi, so the X bar - most probable value - comes
out to be sigma xi divided by n, and this

508
00:46:06,210 --> 00:46:14,560
is nothing but the arithmetic mean. So, what
we proved? We have proved that, if our observations

509
00:46:14,560 --> 00:46:24,330
have only random errors, in that case, the
most probable value of the observations is

510
00:46:24,330 --> 00:46:26,060
the arithmetic mean.

511
00:46:26,060 --> 00:46:35,190
Well, having said that now we will see another
important thing, that is, ‘random error

512
00:46:35,190 --> 00:46:47,040
distribution’. We have already seen some
of the things, but would like to see it in
513
00:46:47,040 --> 00:46:53,480
detail now, how the random errors are actually
distributed, and in doing that for any set

514
00:46:53,480 --> 00:47:00,330
of observations - for example, let us say
this is a variable or a length which we need

515
00:47:00,330 --> 00:47:13,279
to measure. We measure it as xi - i is 1 to
n - and we find the most probable value which

516
00:47:13,280 --> 00:47:19,140
is the arithmetic mean. Also, we can find
one more term that is called sigma, that is,

517
00:47:19,140 --> 00:47:29,640
the standard deviation, which you find by
X bar minus xi - sigma of that - divided by

518
00:47:29,640 --> 00:47:35,759
n minus 1 and a square here - that is called
the standard deviation. What does the standard

519
00:47:35,760 --> 00:47:41,800
deviation show you? The standard deviation
shows you the spread of the point. We will

520
00:47:41,800 --> 00:47:48,370
talk about this standard deviation also later
on, in detail, how we will make use of this.

521
00:47:48,370 --> 00:47:57,130
Well, the random error, the way we are talking
about them, they are actually distributed

522
00:47:57,130 --> 00:48:12,080
in a particular way . Here, it is our variable
X, and this particular value is X bar here,

523
00:48:12,080 --> 00:48:23,140
while here, it is from positive infinity,
minus infinity. Now, this curve - what does

524
00:48:23,140 --> 00:48:31,339
it show? We will talk about it - this is basically
called ‘normal curve’. Also, it is called

525
00:48:31,340 --> 00:48:37,850
Gaussian curve - many times, it is called
‘bell-shaped curve’ because it has the

526
00:48:37,850 --> 00:48:44,310
shape of a bell. Now, what is the meaning
of this? We saw earlier also that our random

527
00:48:44,310 --> 00:48:52,020
error, how they are distributed - just think
of the example of our firing of bullet ; how

528
00:48:52,020 --> 00:48:58,250
those were distributed. If I do it again here,
we are trying to target here, but you end

529
00:48:58,250 --> 00:49:05,860
up firing somewhere here also, but more observations
in this area, more as you go nearer. Then

530
00:49:05,860 --> 00:49:11,000
how the bullets were there; we saw the random
error is distributed. And the same thing is

531
00:49:11,000 --> 00:49:18,740
seen here in this curve. There is an equal
probability of an observation going below

532
00:49:18,740 --> 00:49:28,089
your most probable value on this side - that
means the error is negative - or on the positive

533
00:49:28,090 --> 00:49:33,230
side, the error is positive, because the most
probable value is one which is nearest to

534
00:49:33,230 --> 00:49:39,930
the true value. Any error which is negative
into that will take its value towards negative

535
00:49:39,930 --> 00:49:48,100
side or the less than the most probable value,
and any error which adds to it with the positive

536
00:49:48,100 --> 00:49:53,299
sign will take this value of the variable
towards the positive side.

537
00:49:53,300 --> 00:50:03,970
So, what we see? Our observations or our errors
are equally distributed in both positive and

538
00:50:03,970 --> 00:50:12,330
negative sides; the chances of getting negative
error and positive error are same. Now, here

539
00:50:12,330 --> 00:50:20,470
in this curve, this is our variable, while
this is the probability of occurrence of any

540
00:50:20,470 --> 00:50:27,790
variable. For example, if a variable is here,
x1 - so, what is the probability of occurrence

541
00:50:27,790 --> 00:50:33,400
of x1? The probability of occurrence of x1
is this value. The probability of occurrence

542
00:50:33,400 --> 00:50:41,610
of your most probable value X bar is maximum
- obviously, there is also a probability of

543
00:50:41,610 --> 00:50:48,290
occurring an error which will be infinite
in size, on negative side as well as in positive

544
00:50:48,290 --> 00:50:55,380
side, but its probability will be very less.
So, this is how the curve is: those errors

545
00:50:55,380 --> 00:51:03,770
which are large in magnitude - for example,
here - these errors are large in magnitude,

546
00:51:03,770 --> 00:51:07,570
so their probability - sorry, they are small
in magnitude, and their probability of occurrence

547
00:51:07,570 --> 00:51:13,390
is very high, while the errors which are large
in magnitude on this side, on this part, on

548
00:51:13,390 --> 00:51:18,650
this part - their probability of occurrence
is less. So, this is how estimated random

549
00:51:18,650 --> 00:51:24,090
error is distributed. So, what we will see,
we will look further on this distribution

550
00:51:24,090 --> 00:51:25,710
of the random error.

551
00:51:25,710 --> 00:51:38,470
I am going to draw the same error distribution
or normal curve again, from minus infinite

552
00:51:38,470 --> 00:51:47,709
to plus infinite, and that is our X bar, the
most probable value, and the distribution,

553
00:51:47,710 --> 00:51:58,290
the probability density function of this is
given as ‘f(x)’ which is 1 by sigma and

554
00:51:58,290 --> 00:52:09,279
root 2 pi, e - exponential - raised to power
minus 1/2 into X bar minus x - ‘x’ is

555
00:52:09,280 --> 00:52:17,050
an observation and a variable, while ‘X
bar’ is the most probable value - divided

556
00:52:17,050 --> 00:52:25,740
by sigma square. This is the probability;
f(x) or the Y gives the probability of occurrence

557
00:52:25,740 --> 00:52:31,419
of the variable x. So, this is the equation
for this curve.

558
00:52:33,580 --> 00:52:46,759
Now, what we can do, we can draw the same curve for the error also. This will not
touch

559
00:52:46,760 --> 00:52:57,050
here, they will not meet. Plus infinite, minus
infinite, and in the case of the error, we

560
00:52:57,050 --> 00:53:02,520
will have zero error here, because it is the
most probable value, and the negative values

561
00:53:02,520 --> 00:53:08,190
of the error, and positive values of the error
- and the error will be also distributed in

562
00:53:08,190 --> 00:53:16,400
the similar way, the way the normal distribution
is. And in most of the observations in surveying,

563
00:53:16,400 --> 00:53:22,000
unless the observations are biased because
of some systematic error component, we will

564
00:53:22,000 --> 00:53:27,860
have all our errors generally distributed
in this form.

565
00:53:28,780 --> 00:53:36,360
So, what we saw today, we saw about the errors
in the observation, errors in the measurement,

566
00:53:36,370 --> 00:53:39,620
and this is a very important concept - you
should know about these errors, because in

567
00:53:39,620 --> 00:53:44,290
surveying, you go the field, you measure something
there. So, you should know what kind of errors

568
00:53:44,290 --> 00:53:49,200
may creep in. These errors, they come because of the nature, because of the
instrument,

569
00:53:49,200 --> 00:53:54,750
because of you - the person. Then second thing
we talked about: the types of errors - there

570
00:53:54,750 --> 00:54:00,000
could be mistakes, there could be the systematic
errors; we can eliminate those systematic

571
00:54:00,000 --> 00:54:05,670
errors provided we know the mathematical model
or the physical law which they follow. We
572
00:54:05,670 --> 00:54:11,860
can also do the corrections for all the systematic
errors. Then, after eliminating all the errors,

573
00:54:11,860 --> 00:54:17,510
if still, our observations are not free of
the error, we have another kind of error here,

574
00:54:17,510 --> 00:54:24,280
which is random error. So we saw today a lot
about the random error. What it is; how it

575
00:54:24,280 --> 00:54:31,390
is distributed - it is distributed in a Gaussian
way or the normal curve; we saw the characteristic

576
00:54:31,390 --> 00:54:35,839
of this, that the positive error and the negative
error have got equal likelihood of occurring;

577
00:54:35,840 --> 00:54:40,950
the smaller errors have more probability of
occurring while the large errors have got

578
00:54:40,950 --> 00:54:45,950
less - so, this is what we have seen. What
we will do in our next lecture, we will try

579
00:54:45,950 --> 00:54:52,189
to make use of these concepts and we will
try to, you know, take the normal curve, standard

580
00:54:52,190 --> 00:54:57,940
normal curve, and we will try to make use
of that, so that we understand that what we

581
00:54:57,940 --> 00:55:02,040
have measured, how we represent it, how you present it, how confident we are about
a

582
00:55:02,040 --> 00:55:03,440
particular measurement.

583
00:55:04,540 --> 00:55:05,840
Well, thank you very much.