This is the second appeal in this case involving a boundary line dispute. On March 4, 1981 in an unpublished opinion we reversed that portion of a decree which vested title to a two acre tract in the appellees upon a finding on de novo review that appellant had failed to sustain her burden of proving adverse possession. In that same decree the chancellor found that both parties had failed to establish the common boundary line of their respective properties because of the vast variance between two surveys *17which were introduced and the admissions of both surveyors that neither had tied his survey to an established or known government corner. Both had been unable to reconcile existing monuments set by others, and neither was sure that his survey was a correct one. We remanded the cause and directed that further evidence regarding the location of the true boundary line between the properties be taken.
The parties are adjoining landowners and each has acquired his title by descriptions which make reference to subdivisions which were established in the General Land Office Survey of 1848. Appellant is the owner of the West three-fourths of the Southwest Quarter of the Southwest Quarter of Section 18 and appellees own the East one-quarter of that same Quarter-Quarter. As there has been no adverse possession established by either party, their common boundary lies along the line established in the original land office survey. Only by accurately locating that line could either establish his true boundary.
On remand, testimony of one of the two surveyors who had testified in the first trial was presented. After hearing this evidence the chancellor again found that the appellant had failed in her burden of proving the location of the true boundary line. We agree.
The surveyor testified that his subsequent field and office work convinced him that both surveys used in the prior hearing were in error but it was his opinion that his final survey was an accurate reconstruction of the General Land Office Survey of 1848. He stated that he found no GLO corners in the immediate area of the Southwest Quarter of Section 18 but that he did locate one monument mentioned in the original field notes and survey. Both the plat and survey made reference to a bluff 25 feet high on the west line of the Southwest Quarter of Section 7 (which is immediately north of Section 18). The field notes indicated that the surveyors had placed the west line of Section 7 at a certain point on that bluff which was 220 links west of the break in the bluff which they were required to utilize in an offset survey. From this certain point in the bluff he surveyed north until he found a mound of stones which was within 7 *18feet of the southwest corner of Section 7 set forth in the original field notes. He yielded the distance prescribed in the field notes to that monument and accepted it as the quarter corner of Section 7. In so doing he located the corner at a place other than that established in the field notes of the original survey.
Next, in locating the southwest corner of Section 18 he was required to run south from that starting point a distance of approximately a mile and a half. In the first segment of this subsequent survey he attempted to establish the northwest corner of Section 18. In doing so he did not survey south 1320 feet as directed in the field notes but south 00° 1T 20" west a distance of only 1314.46 feet, yielding the prescribed course and distance to a pile of stones near that point. In doing so he also located this corner at a location other than that established by the original surveyor as shown by his field notes. Although the GLO field notes established the quarter corner of Section 18 at a point south 2629.44 feet from the northwest corner, the surveyor, again “yielding” to another pile of stones, varied that course to south 00° 35' 10" east and the distance to 2657.92 feet. Although the original field notes also ran south 1320 feet to the southwest corner of Section 18 he, again yielding to a pile of stones in the vicinity, followed the course 00° 42' 20" west a distance of 1232.26 feet, again locating the corner at a point different from that at which the field notes of the original survey said it should be.
Not one of the three segments of this survey followed the course or distances set out in the field notes. By varying the distances the southwest corner of Section 18 as located by the surveyor would be approximately 20 feet north of the point at which the original surveyor described it. It is a simple matter of geometric calculation to determine that a variance of 00° 30' west for a distance of 1 mile would place the terminus approximately 41 feet west of a point which would have been reached by the same distance on a course of true south. The one and a half mile segment of this survey varied from the original course by more than half a degree and would place the corner considerably to the west of that prescribed in the field notes. We agree with the learned *19chancellor that this survey did not accurately establish the section line and that the appellant therefore did not meet the burden of proving her boundary.
In varying the courses and distances when yielding to monuments other than those established by the government surveys, the surveyor misunderstood the mission of a surveyor in establishing original government corners. At the time of the Louisiana Purchase this area was a vast wilderness, sparsely populated and with no integrated system of land descriptions. Early in the nineteenth century Congress authorized the General Land Office to survey and plat the entire area in order to give each parcel of land in the public domain a specific and identifiable location which would be subject to relocation by survey at any time. The field notes and plats of those surveys are carefully preserved and with few exceptions all lands in this area were disposed of by the Federal Government with reference to these surveys. Due to errors which were bound to occur, these surveys did not always result in perfect square mile sections of 640 acres with parallel boundaries. Survey parties did not always meet at the point previously calculated and some of the established section corners did not coincide with adjoining ones.
These approved GLO surveys formed the basis for the description of lands when disposed of by the government and any errors, including variances from prescribed courses, distances of acreages, were merged into the government grants. A patent or other original conveyance made with reference to a subdivision conveyed those lands which the General Land Office Plat showed it to contain. Little v. Williams, 88 Ark. 57, 113 S.W. 340 (1908) aff’d 231 U.S. 335 (1913). The purpose of subsequent surveys in locating corners and boundaries is not to correct any error or variance of the original surveyor but is to retrace his steps by use of his field notes and plats and to locate the corners where he located them.
The surveyor testified that he found no corner markers set by the original surveyors because they had long since disappeared. The field notes disclosed that the original *20surveyor marked all corners with wooden stakes and there is no reference to any corner in this vicinity having been marked by rocks. All of the witness monuments except for the possible exception of the 25 foot bluff have been obliterated. Unless the various piles of stones to which this surveyor yielded were shown to have been located where the plat and field notes said they should be, they were of no significance and it was error to yield to them. The Manual of Surveying Instructions published in 1973 by the United States Department of the Interior, Bureau of Land Management, sets out the prescribed method of relocating lost corners. The surveyor testified that he was familiar with that manual and that the methods prescribed by it were accepted as good practice in the surveying profession.
He further testified that, even if his corners had been properly established, good surveying methods would require him to have tied the location of the southeast corner of Section 18 to some known corner to the east of it. He did not do this because he said things in that area were so “messed up” that it would have required an expenditure in excess of $10,000 to accurately locate this property line. The court found from the surveyor’s testimony that depending on the location of a known corner to the east, the location he established could still vary as much as 15 feet in any direction. The chancellor’s finding that a survey containing the possibility of error of this magnitude does not meet that degree of certainty required in the establishment of boundaries between adjoining property owners is not clearly erroneous. Wilson v. Brandenburg, 252 Ark. 921, 481 S.W.2d 715 (1972).
We affirm.
Cooper and Cloninger, JJ., agree.