Chapter 6 Copyright © 1997-2004 Henning Umland All Rights Reserved

Methods for Latitude and Longitude Measurement

Latitude by Polaris

The observed altitude of a star being vertically above the geographic north pole would be numerically equal to the
latitude of the observer (Fig. 6-1).

This is nearly the case with the pole star (Polaris). However, since there is a measurable angular distance between
Polaris and the polar axis of the earth (presently ca. 1°), the altitude of Polaris is a function of the local hour angle. The
altitude of Polaris is also affected by nutation. To obtain the accurate latitude, several corrections have to be applied:

Lat = Ho − 1° + a0 + a1 + a2

The corrections a0, a1, and a2 depend on LHAAries, the observer's estimated latitude, and the number of the month. They
are given in the Polaris Tables of the Nautical Almanac [12]. To extract the data, the observer has to know his
approximate position and the approximate time.

When using a computer almanac instead of the N. A., we can calculate Lat with the following simple procedure. LatE is
our estimated latitude, Dec is the declination of Polaris, and t is the meridian angle of Polaris (calculated from GHA and
our estimated longitude). Hc is the computed altitude, Ho is the observed altitude (see chapter 4).

Hc = arcsin ( sin Lat E ⋅ sin Dec + cos Lat E ⋅ cos Dec ⋅ cos t )

∆ H = Ho − Hc

Adding the altitude difference, ∆H, to the estimated latitude, we obtain the improved latitude:

Lat ≈ Lat E + ∆ H

The error of Lat is smaller than 0.1' when LatE is smaller than 70° and when the error of LatE is smaller than 2°,
provided the exact longitude is known.. In polar regions, the algorithm becomes less accurate. However, the result can
be improved by iteration (substituting Lat for LatE and repeating the calculation). Latitudes greater than 85° should be
avoided because a greater number of iterations might be necassary. The method may lead to erratic results when the
observer is close to the north pole (LatE ≈ DecPolaris). An error in Lat resulting from an error in longitude is not
decreased by iteration. However, this error is always smaller than 1' when the error in longitude is smaller than 1°.
Noon Latitude (Latitude by Maximum Altitude)

This is a very simple method enabling the observer to determine latitude by measuring the maximum altitude of the sun
(or any other object). No accurate time measurement is required. The altitude of the sun passes through a flat maximum
approximately (see noon longitude) at the moment of upper meridian passage (local apparent noon, LAN) when the
GP of the sun has the same longitude as the observer and is either north or south of him, depending on the declination of
the sun and observer’s geographic latitude. The observer’s latitude is easily calculated by forming the algebraic sum or
difference of the declination and observed zenith distance z (90°-Ho) of the sun, depending on whether the sun is north
or south of the observer (Fig. 6-2).

1. Sun south of observer (Fig. 6-2a): Lat = Dec + z = Dec − Ho + 90°

2. Sun north of observer (Fig. 6-2b): Lat = Dec − z = Dec + Ho − 90°

Northern declination is positive, southern negative.

Before starting the observations, we need a rough estimate of our current longitude to know the time (GMT) of meridian
transit. We look up the time of Greenwich meridian transit of the sun on the daily page of the Nautical Almanac and add
4 minutes for each degree of western longitude or subtract 4 minutes for each degree of eastern longitude. To determine
the maximum altitude, we start observing the sun approximately 15 minutes before meridian transit. We follow the
increasing altitude of the sun with the sextant, note the maximum altitude when the sun starts descending again, and
apply the usual corrections.

We look up the declination of the sun at the approximate time (GMT) of local meridian passage on the daily page of the
Nautical Almanac and apply the appropriate formula.

Historically, noon latitude and latitude by Polaris are among the oldest methods of celestial navigation.

Ex-Meridian Sight

Sometimes, it may be impossible to measure the maximum altitude of the sun. For example, the sun may be obscured by
a cloud at this moment. If we have a chance to measure the altitude of the sun a few minutes before or after meridian
transit, we are still able to find our exact latitude by reducing the observed altitude to the meridian altitude, provided we
know our exact longitude (see below) and have an estimate of our latitude. The method is similar to the one used with
the pole star. First, we need the time (UT) of local meridian transit (eastern longitude is positive, western longitude
negative):
Lon [ ° ]
TTransit [ h ] = 12 − EoT [ h ] −
15

The meridian angle of the sun, t, is calculated from the time of observation (GMT):

t [ °] = 15 ⋅ ( TObservation [ h ] − TTransit [ h ] )
Starting with our estimated Latitude, LatE, we calculate the altitude of the sun at the time of observation. We use the
altitude formula from chapter 4:

Hc = arcsin ( sin Lat E ⋅ sin Dec + cos Lat E ⋅ cos Dec ⋅ cos t )

Dec refers to the time of observation. We calculate the difference between observed and calculated altitude:

∆ H = Ho − Hc

We calculate an improved latitude, Latimproved :

Latimproved ≈ Lat E ± ∆ H

(sun north of observer: +∆H, sun south of observer: –∆H)

The exact latitude is obtained by iteration, i. e., we substitute Latimproved for LatE and repeat the calculations until the
obtained latitude is virtually constant. Usually, no more than one or two iterations are necessary. The method has a few
limitations and requires critical judgement. The meridian angle should be small, compared with the zenith distance of
the sun. Otherwise, a greater number of iterations may be necessary. The method may yield erratic results if LatE is
similar to Dec. A sight should be discarded when the observer is not sure if the sun is north or south of his position.

The influence of a longitude error on the latitude thus obtained is not decreased by iteration.

Latitude by two altitudes

Even if no estimated longitude is available, the exact latitude can still be found by observation of two celestial bodies.
The required quantities are Greenwich hour angle, declination, and observed altitude of each body [7].

The calculations are based upon spherical triangles (see chapter 10 & chapter 11). In Fig. 6-3, PN denotes the north
pole, O the observer’s unknown position, GP1 the geographic position of the first body, and GP2 the position of the
second body.
PN

t1

∆GHA

90°-Dec1 90°-Lat 90°-Dec2

O
Fig. 6-3
90°-H1
90°-H2
ψ
ω
ρ
GP1 GP2

First, we consider the spherical triangle [GP1, PN, GP2]. Fig. 6-3 shows only one of several possible configurations. O
may as well be outside the triangle [GP1, PN, GP2]. We form the difference of both Greenwich hour angles, ∆GHA:

∆ GHA = GHA2 − GHA1

Using the law of cosines for sides (chapter 10), we calculate the great circle distance between GP1 and GP2, d.

cos d = sin Dec1 ⋅ sin Dec2 + cos Dec1 ⋅ cos Dec2 ⋅ cos ( ∆ GHA )

d = arccos [ sin Dec1 ⋅ sin Dec2 + cos Dec1 ⋅ cos Dec2 ⋅ cos ( ∆ GHA ) ]

Now we solve the same triangle for the angle ω, the horizontal distance between PN and GP2, measured at GP1:

sin Dec2 − sin Dec1 ⋅ cos d

cos ω =
cos Dec1 ⋅ sin d

 sin Dec2 − sin Dec1 ⋅ cos d 

ω = arccos  
 cos Dec1 ⋅ sin d 

For the spherical triangle [GP1, O, GP2], we calculate the angle ρ, the horizontal distance between O and GP2, measured
at GP1.
sin H 2 − sin H1 ⋅ cos d
cos ρ =
cos H1 ⋅ sin d

 sin H 2 − sin H1 ⋅ cos d 

ρ = arccos  
 cos H1 ⋅ sin d 

We calculate the angle ψ, the horizontal distance between PN and O, measured at GP1. There are two solutions (ψ 1 and
ψ 2) since cos ρ = cos (-ρ):
ψ1 = ω − ρ ψ2 =ω + ρ

The circles of equal altitude intersect each other at two points. The corresponding positions are on opposite sides of the
great circle going through GP1 and GP2 (not shown in Fig. 6-3). Using the law of cosines for sides again, we solve the
spherical triangle [GP1, PN, O] for Lat. Since we have two solutions for ψ, we obtain two possible latitudes, Lat1 and
Lat2.
sin Lat1 = sin H1 ⋅ sin Dec1 + cos H1 ⋅ cos Dec1 ⋅ cosψ 1

Lat1 = arcsin (sin H1 ⋅ sin Dec1 + cos H1 ⋅ cos Dec1 ⋅ cosψ 1 )

sin Lat 2 = sin H1 ⋅ sin Dec1 + cos H1 ⋅ cos Dec1 ⋅ cosψ 2

Lat 2 = arcsin (sin H1 ⋅ sin Dec1 + cos H1 ⋅ cos Dec1 ⋅ cosψ 2 )

We choose the value nearest to our estimated latitude. The other one is discarded. If both solutions are very similar and
a clear distinction is not possible, one of the sights should be discarded, and a body with a more favorable position
should be chosen.

Although the method requires more complicated calculations than, e. g., a latitude by Polaris, it has the advantage that
measuring two altitudes usually takes less time than finding the maximum altitude of a single body. Moreover, if fixed
stars are observed, even a chronometer error of several hours has no significant influence on the resulting latitude since
∆GHA and both declinations change very slowly in this case.

When the horizontal distance between the observed bodies is in the vicinity of 0° or 180°, the observer's position is
close to the great circle going through GP1and GP2. In this case, the two solutions for latitude are similar, and finding
which one corresponds with the actual latitude may be difficult (depending on the quality of the estimate). The resulting
latitudes are also close to each other when the observed bodies have approximately the same Greenwich hour angle.
Noon Longitude (Longitude by Equal Altitudes)

Since the earth rotates with an angular velocity of 15° per hour with respect to the mean sun, the time of local meridian
transit (local apparent noon) of the sun, TTransit, can be used to calculate the observer's longitude:

Lon [ ° ] = 15 ⋅ ( 12 − TTransit [ h ] − EoTTransit [ h ] )

TTransit is measured as GMT (decimal format). The correction for EoT at the time of meridian transit, EoTTransit, has to be
made because the apparent sun, not the mean sun, is observed (see chapter 3). Since the Nautical Almanac contains only
values for EoT (see chapter 3) at 0:00 GMT and 12:00 GMT of each day, EoTTransit has to be found by interpolation.

Since the altitude of the sun - like the altitude of any celestial body - passes through a rather flat maximum, the time of
peak altitude is difficult to measure. The exact time of meridian transit can be derived, however, from the times of two
equal altitudes of the sun.

Assuming that the sun moves along a symmetrical arc in the sky, TTransit is the mean of the times corresponding with a
chosen pair of equal altitudes of the sun, one occurring before LAN (T1), the other past LAN (T2) (Fig. 6-4):

T1 + T2
TTransit =
2

In practice, the times of equal altitudes of the sun are measured as follows:

In the morning, the observer records the time T1 corresponding with a chosen altitude, H. In the afternoon, the time T2 is
recorded when the descending sun passes through the same altitude again. Since only times of equal altitudes are
measured, no altitude correction is required. The interval T2-T1 should be greater than approx. 2 hours.
Unfortunately, the arc of the sun is only symmetrical with respect to TTransit if the sun's declination is constant during the
observation interval. This is approximately the case around the times of the solstices. During the rest of the year,
particularly at the times of the equinoxes, TTransit differs significantly from the mean of T1 and T2 due to the changing
declination of the sun. Fig. 6-5 shows the altitude of the sun as a function of time and illustrates how the changing
declination affects the apparent path of the sun in the sky, resulting in a time difference, ∆T.

The blue line shows the path of the sun for a given, constant declination, Dec1. The red line shows how the path would
look with a different declination, Dec2. In both cases, the apparent path of the sun is symmetrical with respect to TTransit.
However, if the sun's declination varies from Dec1 at T1 to Dec2 at T2, the path shown by the green line will result.
Now, T1 and T2 are no longer symmetrical to TTransit. The sun's meridian transit occurs before (T1+T2)/2 if the sun's
declination changes toward the observer's parallel of latitude, like shown in Fig. 6-5. Otherwise, the meridian transit
occurs after (T1+T2)/2. Since time and local hour angle (or meridian angle) are proportional to each other, a systematic
error in longitude results.

The error in longitude is negligible around the times of the solstices when Dec is almost constant, and is greatest (up to
several arcminutes) at the times of the equinoxes when the rate of change of Dec is greatest (approx. 1'/h). Moreover, the
error in longitude increases with the observer's latitude and may be quite dramatic in polar regions.

The obtained longitude can be improved, if necessary, by application of the equation of equal altitudes [5]:

 tan Lat tan Dec2 

∆ t ≈  −  ⋅ ∆ Dec ∆ Dec = Dec 2 − Dec1
 sin t 2 tan t 2 

∆t is the change in the meridian angle, t, which cancels the change in altitude resulting from a small change in
declination, ∆Dec. Lat is the observer's latitude. If the accurate latitude is not known, an estimated latitude may be used.
t2 is the meridian angle of the sun at T2. Since we do not know the exact value for t2 initially, we start our calculations
with an approximate value calculated from T1 and T2:

15 ⋅ ( T2 [ h ] − T1 [ h ] )
t2 [ ° ] ≈
2
We denote the improved value for T2 by T2*.

∆ t [° ]
T2*[ h ] = T2 [ h ] − ∆T [ h ] = T2 [ h ] −
15

At T2*, the sun would pass through the same altitude as measured at T1 if Dec did not change during the interval of
observation. Accordingly, the improved time of meridian transit is:

T1 + T2*
TTransit =
2
The residual error resulting from the initial error of t2 is usually not significant. It can be decreased, if necessary, by
iteration. Substituting T2* for T2, we get the improved meridian angle, t2*:

t 2* [ ° ] ≈
(
15 ⋅ T2* [ h ] − T1 [ h ] )
2
With the improved meridian angle t2*, we calculate the improved correction ∆t*:

 tan Lat tan Dec2 

∆ t * ≈  −  ⋅ ∆ Dec

 sin t *
2 tan t *
2 

Finally, we obtain a more accurate time value, T2**:

∆ t* [ ° ]
T2* *[ h ] = T2 [ h ] −
15

And, accordingly:

T1 + T2* *
TTransit =
2
The error of ∆Dec should be as small as possible. Calculating ∆Dec with a high-precision computer almanac is
preferable to extracting it from the Nautical Almanac. When using the Nautical Almanac, ∆Dec should be calculated
from the daily change of declination to keep the rounding error as small as possible.

Although the equation of equal altitudes is strictly valid only for an infinitesimal change of Dec, dDec, it can be used for
a measurable change, ∆Dec, (up to several arcminutes) as well without sacrificing much accuracy. Accurate time
measurement provided, the residual error in longitude rarely exceeds ±0.1'.

Theory of the equation of equal altitudes

The equation of equal altitudes is derived from the altitude formula (see chapter 4) using differential calculus:

sin H = sin Lat ⋅ sin Dec + cos Lat ⋅ cos Dec ⋅ cos t

First, we need to know how a small change in declination would affect sin H. We form the partial derivative with respect
to Dec:

∂ ( sin H )
= sin Lat ⋅ cos Dec − cos Lat ⋅ sin Dec ⋅ cos t
∂ Dec

Thus, the change in sin H caused by an infinitesimal change in declination, d Dec, is:

∂ ( sin H )
⋅ d Dec = ( sin Lat ⋅ cos Dec − cos Lat ⋅ sin Dec ⋅ cos t ) ⋅ d Dec
∂ Dec

Now, we form the partial derivative with respect to t in order to find out how a small change in the meridian angle would
affect sin H:

∂ ( sin H )
= − cos Lat ⋅ cos Dec ⋅ sin t
∂t

The change in sin H caused by an infinitesimal change in the meridian angle, dt, is:

∂ ( sin H )
⋅ d t = − cos Lat ⋅ cos Dec ⋅ sin t ⋅ d t
∂t

Since we want both effects to cancel each other, the total differential has to be zero:

∂ ( sin H ) ∂ ( sin H )
⋅ d Dec + ⋅d t = 0
∂ Dec ∂t

∂ ( sin H ) ∂ ( sin H )
− ⋅d t = ⋅ d Dec
∂t ∂ Dec
 cos Lat ⋅ cos Dec ⋅ sin t ⋅ d t = ( sin Lat ⋅ cos Dec − cos Lat ⋅ sin Dec ⋅ cos t ) ⋅ d Dec

sin Lat ⋅ cos Dec − cos Lat ⋅ sin Dec ⋅ cos t

dt = ⋅ d Dec
cos Lat ⋅ cos Dec ⋅ sin t

 tan Lat tan Dec 

d t =  −  ⋅ d Dec
 sin t tan t 

 tan Lat tan Dec 

∆ t ≈  −  ⋅ ∆ Dec
 sin t tan t 

Longitude Measurement on a Traveling Vessel

On a traveling vessel, we have to take into account not only the influence of varying declination but also the effects of
changing latitude and longitude on the altitude of the body during the observation interval. Differentiating sin H (altitude
formula) with respect to Lat, we get:

∂ ( sin H )
= cos Lat ⋅ sin Dec − sin Lat ⋅ cos Dec ⋅ cos t
∂ Lat

Again, the total differential is zero because the combined effects of latitude and meridian angle cancel each other with
respect to their influence on sin H:

∂ ( sin H ) ∂ ( sin H )
⋅ d Lat + ⋅ dt = 0
∂ Lat ∂ Dec

In analogy with a change in declination, we obtain the following formula for a small change in latitude:

 tan Dec tan Lat 

d t =  −  ⋅ d Lat
 sin t tan t 

The correction for the combined variations in Dec, Lat, and Lon is:

 tan Lat 2 tan Dec2   tan Dec2 tan Lat 2 

∆ t ≈  −  ⋅ ∆ Dec +  −  ⋅ ∆ Lat − ∆ Lon
 sin t 2 tan t 2   sin t 2 tan t 2 

∆Lat and ∆Lon are the small changes in latitude and longitude corresponding with the path of the vessel traveled
between T1 and T2. The meridian angle, t2, has to include a correction for ∆Lon:

15 ⋅ ( T2 [ h ] − T1 [ h ] ) − ∆ Lon [ ° ]
t2 [ ° ] ≈
2

∆Lat and ∆Lon are calculated from the course, C, the velocity over ground, v, and the time elapsed.

∆ Lat [ ' ] = v [ kn ] ⋅ cos C ⋅ ( T2 [ h ] − T1 [ h ] )

 Lat 2 = Lat1 + ∆ Lat

sin C
∆ Lon [ ' ] = v [ kn ] ⋅ ⋅ ( T2 [ h ] − T1 [ h ] )
cos Lat

Lon2 = Lon1 + ∆ Lon

1 kn ( knot ) = 1 nm / h

C is measured clockwise from true north (0°...360°). Again, the corrected time of equal altitude is:

∆t [° ]
T2* [ h ] = T2 [ h ] −
15

T1 + T2*
TTransit =
2

The longitude calculated from TTransit refers to the observer's position at T1. The longitude at T2 is Lon+∆Lon.
The longitude error caused by a change in latitude can be dramatic and requires the navigator's particular attention, even
if the vessel travels at a moderate speed. The above considerations clearly demonstrate that determining one's exact
longitude by equal altitudes of the sun is not as simple as it seems to be at first glance, particularly on a traveling vessel.
It is therefore quite natural that with the development of position line navigation (including simple graphic solutions for
a traveling vessel), longitude by equal altitudes became less important.

The Meridian Angle of the Sun at Maximum Altitude

Fig. 6-5 shows that the maximum altitude of the sun is slightly different from the altitude at the moment of meridian
passage if the declination changes. At maximum altitude, the rate of change of altitude caused by the changing
declination cancels the rate of change of altitude caused by the changing meridian angle.
The equation of equal altitude enables us to calculate the meridian angle of the sun at this moment. We divide each side
of the equation by the infinitesimal time interval dT:

dt  tan Lat tan Dec  d Dec

=  − ⋅
dT  sin t tan t  dT

Measuring the rate of change of t and Dec in arcminutes per hour we get:

 tan Lat tan Dec  d Dec [ ' ]

900 ' / h =  − ⋅
 sin t tan t  d T [ h ]

Since t is a very small angle, we can substitute tan t for sin t:

tan Lat − tan Dec d Dec [ ' ]

900 ≈ ⋅
tan t dT [h ]

Now, we can solve the equation for tan t:

tan Lat − tan Dec d Dec [ ' ]

tan t ≈ ⋅
900 dT [h ]
Since a small angle (in radians) is nearly equal to its tangent, we get:

π tan Lat − tan Dec d Dec [ ' ]

t [° ] ⋅ ≈ ⋅
180 900 dT [h ]

Measuring t in arcminutes, the equation is stated as:

d Dec [ ' ]
t [ ' ] ≈ 3.82 ⋅ ( tan Lat − tan Dec ) ⋅
dT [h ]

dDec/dT is the rate of change of declination measured in arcminutes per hour.

The maximum altitude occurs after meridian transit if t is positive, and before meridian transit if t is negative.

For example, at the time of the spring equinox (Dec ≈ 0, dDec/dT ≈ +1'/h) an observer being at +80° (N) latitude would
observe the maximum altitude of the sun at t ≈ +21.7', i. e., 86.8 seconds after meridian transit (LAN). An observer at
+45° latitude, however, would observe the maximum altitude at t ≈ +3.82', i. e., only 15.3 seconds after meridian transit.

The Maximum Altitude of the Sun

We can use the last equation to evaluate the systematic error of a noon latitude. The latter is based upon the maximum
altitude of the sun, not on the altitude at the moment of meridian transit. Following the above example, the observer at
80° latitude would observe the maximum altitude 86.7 seconds after meridian transit.

During this interval, the declination of the sun would have changed from 0 to +1.445'' (assuming that Dec is 0 at the time
of meridian transit). Using the altitude formula (chapter 4), we get:

Hc = arcsin ( sin 80° ⋅ sin 1.445' ' + cos 80° ⋅ cos 1.445' ' ⋅ cos 21.7') = 10° 0' 0.72' '

In contrast, the calculated altitude at meridian transit would be exactly 10°. Thus, the error of the noon latitude would be
-0.72''.
In the same way, we can calculate the maximum altitude of the sun observed at 45° latitude:

Hc = arcsin ( sin 45° ⋅ sin 0.255' ' + cos 45° ⋅ cos 0.255' ' ⋅ cos 3.82') = 45° 0' 0.13' '

In this case, the error of the noon latitude would be only -0.13''.
The above examples show that even at the times of the equinoxes, the systematic error of a noon latitude caused by the
changing declination of the sun is not significant because it is much smaller than other observational errors, e. g., the
errors in dip or refraction. A measurable error in latitude can only occur if the observer is very close to one of the poles
(tan Lat!). Around the times of the solstices, the error in latitude is practically non-existent.
Time Sight

The process of deriving the longitude from a single altitude of a body (as well as the observation made for this purpose)
is called time sight. However, this method requires knowledge of the exact latitude, e. g., a noon latitude. Solving the
navigational triangle (chapter 11) for the meridian angle, t, we get:

sin Ho − sin Lat ⋅ sin Dec

t = ± arccos
cos Lat ⋅ cos Dec

The equation has two solutions, +t and –t, since cos t = cos (–t). Geometrically, this corresponds with the fact that the
circle of equal altitude intersects the parallel of latitude at two points.
Using the following formulas and rules, we obtain the longitudes of these points of intersection, Lon1 and Lon2:

Lon1 = t − GHA

Lon2 = 360° − t − GHA

If Lon1 < − 180° → Lon1 + 360°

If Lon2 < − 180° → Lon2 + 360°
If Lon2 > + 180° → Lon2 − 360°

A time sight can be used to derive a line of position from a single assumed latitude. After solving the time sight, we plot
the assumed parallel of latitude and the calculated meridian.

Next, we calculate the azimuth of the body with respect to the position thus obtained (azimuth formulas, chapter 4) and
plot the azimuth line. Our line of position is the perpendicular of the azimuth line going through the calculated position
(Fig. 6-6).

The latter method is of historical interest only. The modern navigator will certainly prefer the intercept method (chapter
4) which can be used without any restrictions regarding meridian angle (local hour angle), latitude, and declination (see
below).

A time sight is not reliable when the body is close to the meridian. Using differential calculus, we can demonstrate that
the error of the meridian angle, dt, resulting from an altitude error, dH, varies in proportion with 1/sin t:

cos Ho
dt = − ⋅ dH
cos Lat ⋅ cos Dec ⋅ sin t

Moreover, dt varies inversely with cos Lat and cos Dec. Therefore, high latitudes and declinations should be avoided as
well. The same restrictions apply to Sumner's method.

Direct Computation of Position

Combining a time sight with a latitude by two altitudes, we can find our position, provided we know the exact time.
After obtaining our latitude, Lat, from two altitudes (see above), we apply the time sight formula and calculate the
meridian angle of the first body, t1, from the quantities Lat, Dec1, and H1 (see Fig. 6-3). From t1, we obtain two possible
longitudes. We choose the one nearest to our estimated longitude. This is a rigorous method, not an approximation. It is
rarely used, however, since it is more cumbersome than the graphic solutions described in chapter 4.