delivered the opinion of the court.

On July 3, 1913, the District Court of Ag’uadilla rendered a judgment in favor of the complainants in this action. This was reversed on appeal mainly because the proof was insufficient, a filiation suit requiring’ strong proof. Méndez v. Martinez, 21 P. R. R. 238. The case went back for a new trial and the court again found in favor of the complainants. Prom this second judgment Víctor P. Martínez has again appealed.

A number of the errors assigned arose from the situation which the appellant deemed to have been produced by the facts developed at the first trial.

The third exception presented by the appellant is of this nature. The complainants offered the testimony of Cecilia Mendez, Sixta Torres, and Irene Hernández,. all of whom had testified at the first. trial. When they were put upon the stand and in advance of any question directed to them the appellant objected to their testimony on the theory that it was first necessary to justify the way or manner in which the testimony about to be given would differ from the facts developed at the first trial. He seems to have the idea that when a case is sent back for a new trial the facts of the first trial are a kind of res adjudicata and that only new evidence *227or newly discovered evidence could be offered, and that in the second trial only additional or explanatory facts should be permitted. We shall not discuss all the misapprehensions of the appellant. It is enough to say that when a case is sent back for a new trial in general terms to permit the clearing up of all the main issues, as was done here, the case is a trial de novo and the parties are no more limited in their proof than they would be on an appeal from a municipal court or if they dismissed their case and began again. On certain occasions when this court reverses for lack of the elements of proof or for an accounting the order to be followed by the court below is indicated by our judgment. In the absence of a specific indication to the contrary, however, the ordering of a new trial leaves the parties free to begin again with no more let or hindrance than that the witnesses may be confronted with inconsistent statements made at the former trial. The principles governing motions for new trials on the ground of newly discovered evidence have no application to a case reversed for errors in the admission or rejection of proof, or in the lack of sufficient evidence or the like. Donahue v. Klassner, 22 Mich. 252; Atchison v. Owen, 58 Tex. 610; Jones v. Andrews, 72 Tex. 5; 29 Cyc. 1043-44. Of some application is the decision of this court in the case of Carmona et al. v. Cuesta, 23 P. R. R. 636.

The fourth exception of the appellant was based on the same erroneous theory. Of the same nature are one or more of the reasons assigned in a motion of nonsuit, to the refusal of which the fifth exception relates.

The court refused to admit in evidence the statement of facts of the first trial, and to this refusal the seventh exception relates. The offering of this statement of facts as independent evidence is none of the modes recognized by the law where the witnesses themselves are living and may be brought before the court. The rule is that the witnesses themselves must be produced, and the exceptions are based upon the absence or inability of such witnesses to testify.

*228So far as this offer of the testimony at the former trial was an attempt to contradict the statements of the witnesses made at the second trial, it is perhaps sufficient to say that the appropriate mode of contradiction is set forth in sections 156 and 157 of the Law of Evidence. The witnesses must be testifying, their inconsistent statements must be drawn to their attention, and they must have an opportunity to explain. In another of his exceptions, in the assignment of error that relates to the admission of the letters of the appellant’s father, the appellant also had the idea that he could attack such admissibility by the presentation of the facts of the first trial.

The appellant offered in evidence the stenographic report of the first trial to contradict certain witnesses and the court rejected the same, and this was made the subject of the ninth exception. While this alleged error is elaborately discussed in the briefs of the parties, it seems sufficient to say that the witnesses sought to be impeached were not confronted with their inconsistent statements nor given an opportunity to explain, and the exception is met by the principles of sections 156, 157 and 158 of the Law of Evidence, supra.

The complainant, to prove the birth of the complainants, offered in evidence the two certificates issued by the clexdr of the Civil Registry of San Sebastián. In one of these certificates Cecilia Méndez is made to appear and say that a child was born to her some twenty-nine days before. The law seems to require that such an appearance must be made within twenty days, although the appellant says the period is eight days. The particular period makes no difference because the appearance of Cecilia Méndez was' outside of either period, and, hence, the clerk of the registry was without authority to make the entry. We are inclined to think that the admission of this certificate was error. Generally, when a record is relied on to dispense with the necessity of other evidence the record must be shown to come strictly within the requirements of the law. To the other certificate *229tile objection was made that, in the absence of. explanation why one of the parents did not appear, the midwife had no right to appear. We likewise think the objection to the second certificate was a valid one. The errors, however, in the admission of this evidence were entirely harmless. There was scarcely a .doubt that these children were the children of Cecilia Méndez. In a suit brought by these children against their mother to claim filiation the proof might have had more importance, but in the case at bar there was no serious question of the fact that the complainants were the children of Cecilia Méndez and that they were born on the dates specified. The relation of these children to Celilia Méndez was abundantly proved by her own testimony, by the testimony of her other children, by the midwife who assisted at their birth, and by other witnesses.

The second exception relates to the admission by the court of numerous letters purporting to have been written by Victor Martínez y Martinez to Cecilia Méndez. A part of .the argument goes to the lack of identification. There was ample proof tending to show that these letters were written and sent by the father of the appellant. Most of the argument, however, goes to the weight of the proof and should be discussed under the sufficiency of the proof to sustain the judgment. It would be an objection to say of these letters that they related to a period outside of the issues in this case. The appellant’s theory for this purpose is that the complaint relates' only to the period existing between the conception of the children and the death of the natural father, namely, from 1908 to 1912; but the theory of the complaint is that the relations between Víctor Martínez y Martínez and Cecilia Méndez existed during numerous years, hence it was relevant and material, if not necessary, for the complainants to show a continuous state of concubinage extending perhaps even beyond an earlier period than the conception of the complainants Rven if the relations were inter-*230rapted, the admission of several letters outside of the dates covered by the issues in this case would be harmless in view of the numerous letters properly admissible within these elates, unless the appellant could show that the letters outside of the period were of a very different nature from the letters written during the particular period of the issues covered by this case. As we have indicated, these objections go rather to the weight of the evidence. Some of the other objections to these letters we have considered elsewhere.

At the close of the complainants’ testimony the defendant presented a motion for nonsuit. A motion for a nonsuit must necessarily be directed to the weight of the evidence. Some of the matters raised by this motion for nonsuit we have already discussed, but it is sufficient to say with regard to motions for nonsuit in general that when a defendant goes on and presents his evidence he waives sucli motion. Rivera v. Díaz, 19 P. R. R. 524; Príncipe v. The American R. R. Co., 22 P. R. R. 282. This court, then, will only examine the whole evidence to see whether the facts support the judgment.

Appellant complains of the refusal of the court to admit a certificate of death to show the age of his father. The court admitted the certificate for the purpose of showing the death, but excluded it for the purpose of showing the age and we find no error in its action.

The eighth exception relates to the refusal of the court to admit two certificates of the municipal court showing that Joaquin Oronoz had filed two complaints against the appellant for supposed crimes, of which the appellant was acquitted. These certificates did not tend to show that Joaquin Oronoz had any interest in this suit, and the fact that the witness said that he had only filed one complaint against the defendant was a statement collateral to the issuer of the case. If the witness is asked about a collateral matter having no relation to the facts of the trial, evidence to show that he made a false statement is generally inadmissible for *231tlie reason that if such collateral inquiries were permitted a trial might never come to an end'. Wigmore on Evidence,, pars. 1001 et seq.; 40 Cyc. 2569 et seq.; sections 156,157, 158, of the Law of Evidence. The proof, moreover, tended to show in any event that the supposed misstatement was not littered corruptly.

The tenth exception turned upon the refusal of the court to admit a letter wherein Víctor Martinez, Senior, wrote a letter to his wife in which he said that Víctor P. Martinez was his only legitimate son. This letter was in the nature of a self-serving declaration. It was res inter alios acta and no reason was shown for its admission. We find no error and the letter would have practically no probative force of any kind. Without the solemnity of an oath a letter written by a man to his wife in which he says he has no other child but his legitimate son, even if it were competent evidence, would be even less convincing or probatory than ■ the statement of a man that a piece of property was his own and not ganancial. This disposes of all the exceptions presented in the record.

The appellant in his brief argues the insufficiency of both the complaint and the judgment. We think the complaint sufficiently sets forth that at the time of the conception and birth of each of the complainants in this'case, Víctor Martinez and Cecilia Méndez lived together in a state of concubinage, having had amorous relations, and that at the time-of the conception of the children there was no impediment to the marriage of their parents. The complaint also sets forth facts which tend to show that Víctor Martinez had always considered the children as his own and had called them so on various different occasions, and that Victor Martinez was dead. The complaint, among other things, asks that the children be declared the natural acknowledged children of Víctor Martinez, with the right to' bear his name and the other rights which the Civil Code gave them; and the judgment in this case sufficiently followed the complaint. *232It is no criticism of the judgment that it declared the civil rights the children had to the inheritance of their father.

The really important matters to he considered in this case are whether the proof was strong enough to justify the judgment of the court, and also whether such proof gave the children a legal status.

In reversing the judgment in the first appeal this court deviated from its general practice of rendering the judgment that the court below ought to have rendered, because it seemed probable that the complainants failed to present all the evidence available to them. This court found that the acts of Víctor Martinez in caressing his alleged children and sometimes calling them his children were not the strong proof which the law required, and were rather dubious ways to .show the intent of the father to acknowledge his children. We also found that the letters offered in that case were not sufficiently identified and were not shown to cover the period immediately before and after the birth of the complainants in this case. The theory maintained by the parties at the previous trial was that the alleged parents of these children had maintained relations for only four or five years, and, hence, that as the letters covered a period anterior to that, they would have no probatory force to show a relation existing after the conclusion of the correspondence. There was no proof that any of these letters related to the period of the conception and. birth of the complainants in this case. The court also found that the statements of the witnesses with regard to the support of Cecilia Méndez and her children by Víctor Martinez were unconvincing and even contradictory.

At the second trial Cecilia Méndez took the stand and testified that she • had relations with Víctor Martinez for twenty-three years; that Victor Martinez during the whole period generally came to her house about once a week and stayed there three or four days; that during these three or four days when he was with her he led the life of an *233-ordinary man with, a wife and children; that lie looked after lier and supported her and supported all her children; that he sent her eight dollars a week; that whatever property she had had come from him; that he maintained her in various different places and helped her to maintain and manage these places; that his relations with her continued- up to and after the birth of Pedro Angel and Laura Maria, the complainants- in this case; that at the time of the trial one of these children was five years old and the other was three, and in this way she identified the birth of the children as being at the time mentioned in the complaint, namely, November 22, 1909, and November 4, 1911; that she knew that Víctor Martinez was married and that he had a house elsewhere, sometimes in San Sebastián and sometimes in Maya-giiez. She gave other evidence tending to show that during the whole twenty-three years Víctor Martínez' came to see her periodically and lived a marital life with her for three or four days of each of his visits. There was undisputed proof in this case that Víctor Martínez y Martinez had an independent house wherein he lived with his wife and that his only legitimate child was the defendant.

Cecilia Méndez is the person in whose possession all the facts of the case would naturally be. She is impugned on the ground of interest, and as a general thing it would be dangerous for any court of first instance to take the uncorroborated statements of a woman in her condition, but Cecilia Méndez in this case was strongly corroborated. She was corroborated by her other children, concerning whom there was also proof that they were the children of Victor Martinez; she was corroborated by witnesses who were present at the time of the birth of her last two children, by witnesses who testified to their knowledge of the long-continued relations between Víctor Martínez and Cecilia Mén-dez, by witnesses who bore letters and money to her and, finally, not to mention all the circumstances, by the letters themselves.

*234We have examined the signatures to these letters. The appellant is right in maintaining that the signature to the letters differs from the signature to the unquestioned legal documents. The legal documents were of a date slightly anterior to the dates of the first letters. We think, nevertheless, that the general character of the signatures is the-same. The individual letters or characters forming the-signatures to the correspondence differ no more from the-letters in the unquestioned documents than they do from, the individual characters in the body of the correspondence.. In some of the letters admitted in evidence the ‘ ‘ M ’ ’ in Martinez differs from the “M” in the body of the communication. The letters at this trial were identified by various-witnesses, by the persons who bore them, and by the members of the family of Cecilia Méndez who had seen or received them. The contents of these letters tended to prove-a continuous relation between Víctor Martínez and Cecilia Méndez. One of these letters was a letter written by one-of Cecilia Mendez’ sons to his mother while he was with his-alleged putative father and the putative father adds a few lines to the words of the boy. This letter was attacked because it was signed “Pascasio” when the boy’s name-was Francisco, but we think there was sufficient evidence that the boy was called by both names. Now, while these-letters are not clearly shown to extend over to the period in which the complainants were born, yet all the proof tends to show that the relations that existed during all the period of the letters continued to exist thereafter and that the status- or condition, if it may so be called, existing between Victor Martínez and Cecilia Méndez persisted during the conception and birth of Pedro Angel and Laura Maria.

The case presented at the second trial was very different from the meagre case presented at the first trial. In the-absence of any real indicia of fraud or mendacity the mere-fact that the testimony of the witnesses differed from that, given at the first trial in parts of their testimony and that *235tlie first trial-court theory was somewhat changed do not establish the fact that the witnesses were not telling* the truth. If there was any substantial difference in the testimony, the complainants were entitled to the real truth and the court was bound to find it.

The record does not show on what ground the court below based its conclusions that these complainants had a right to compel recognition. We should be inclined to doubt whether there was any sufficient proof that Víctor Martínez y Martinez really acknowledged these particular two children as his own. There was proof, of course, tending to that effect* but we doubt whether we should be satisfied with that proof alone. We have omitted to say that the wife of Victor Martinez died in 1908 and hence that thereafter there was no impediment to the marriage of Victor Martinez and Cecilia Méndez. The rights of these children to be called natural acknowledged children depend, among other things, on the capacity of their parents to contract marriage. We might have some slight doubt as to whether Víctor Martinez led a family life with respect to these particular children, as the letters do not cover the period beginning with the conception of the children, but that doubt is involved in what we consider to be the principal question in this case, namely* whether the acts of Víctor Martinez with Cecilia Méndez constituted a state of concubinage sufficiently known to-others.

Paragraph 3 of section 189 of the Revised Civil Code* before the amendment of March 9, 1911, provided that a father was obliged to recognize his illegitimate child when the mother was known to have lived in concubinage during the pregnancy or at the time of the birth of the child or when the child was born while his parents were engaged to be married.

And section 193 of the Civil Code, as amended by Act No. 73 of March 9, 1911, provides that the father is obliged to acknowledge 'the natural child “When the mother was, *236known to liave lived in concubinage with the father, both during her pregnancy and at the time of the birth of the child.”

Some point .is made of the fact that Pedro Angel was born before March 9, 1911, and Laura Maria was born after-wards. However, the only way in which we would be satisfied to affirm the judgment in this case is on the theory that the facts showed the state of concubinage described- in section 3 of each law. We shall lay no stress in this consideration on whether this concubinage was known. We think the proof shows that the relations between Víctor Martínez and Cecilia Mendez were sufficiently known in San Sebastián, where the man and the woman were seen to be living together.

Escriche in Volume II, page 453, defines a concubine as one who lives with a man as if he were her husband, when both are free and able to contract marriage, although the man may not, in fact, be her husband, and Gregorio López, the commentator on the Laws of the Partidas, expresses himself similarly.

Article 452 of the Penal Code formerly in force in Cuba and Porto Bic.o punished the husband who had a concubine Inside his house or notoriously outside of his house. .These citations indicate by themselves that it is not a necessary element of a state of concubinage that a man must take up a more or less exclusive residence with the woman, as the appellant maintains. We do not think that it is necessary for a state of concubinage to exist that a man should have no other residence than that of the woman. As was pointed nut by the Supreme Court of Louisiana, such a condition is not indispensable in a state of matrimony, nor was it for •a state of concubinage. Succession of Jahraus, infra. The .Enciclopedia Jurídica Española has a short article on con-uubinage, where it is expressed that the union that exists between a man and a woman to constitute concubinage should be enduring, continuous and persistent.

*237Louisiana, wliicli also indirectly inherits its law from. Borne, also had a provision cf law similar to section 452 of the Penal Code for Spain and Cuba, where it is made a crime for a husband to keep a concubine in the matrimonial dwelling or openly and publicly in any other. The statute is set forth in the case of Ledoux v. Her Husband, 10 La. Ann. 663.

Section 1481 of the Civil Code of Louisiana provides that those who have lived together in open concubinage are incapable of making a donation of immovables one to the other.

In the case of Succession of Jahraus, 114 La. 456, a married man maintained a woman secretly and hence the court decided that section 1481 did not apply, but in the course of the opinion the court said:

“Defendant denies that illicit relations ever existed between him and the testatrix, and, in case the court should find that such relations did exist, he denies that they were of the character described, or, in other .words, that they constituted the ‘living together in open concubinage’ from which results the incapacity pronounced by article 1481.

#

“Defendant’s learned counsel base an argument upon the words ‘have lived together.’ Words', they say, must be taken in their ordinary sense in the interpretation of a statute, and the ordinary sense of the words ‘have lived together’ is ‘have dwelt or resided together’; and therefore the concubines must have lived together in the sense of dwelt or resided together, in order that article 1481 should apply to them.

“We cannot adopt that construction. Residing together is a frequent concomitant of concubinage, but it is not an essential feature. There may be concubinage, and open concubinage at that, without it, just as there may be marriage without it. It is an invariable concomitant of marriage, but not an essential; its absence would not invalidate the marriage. We do not think it would do to say that the law will take cognizance of a concubinage so long as the concubines abstain from actually residing together. What the law aims at is the relation, the permanent relation of living together as man and wife Avitliout being married, and, if that relation is main*238tained openly, the condition of article 1481 is fulfilled, even though the parties do not reside together. Under the interpretation contended for, a man might set up an establishment for a woman, visit her there regularly, raise a family with her, pay her bill?, educate her children, by word or conduct, or both, avow his illicit relations with her, and yet the case not come within the purview of article 1481 so long as he resided elsewhere. We repeat, we do not think it would do to put that interpretation upon article 1481.”

Tn the case of Succession of Filhiol, 119 La. 998, the court distinguished the case of Jahraus with respect to what constituted a secret relation between the parties and found a state of concubinage to exist between them where the facts, barring the question of being notoriously known, were very similar to the facts of the case at bar. The reasoning of the court in the case of Jahraus that we have quoted was accepted.

In the case of Jahraus, supra, on page 49, a French authority is quoted, defining concubinage to a similar effect as that defined in said case. From the antecedents in Porto Bico, and the law we have been able to find in the jurisdictions having conditions similar to Porto Bico, we have no hesitation in concluding that when a man maintains a separate establishment for a woman over a period of years and where he goes to visit her, staying with her three or four days at a time, such a relation constitutes a state of .concubinage.

Appellant attacks the judgment and general finding of the court below on account of prejudice, but his argument is principally dependent upon the fact that the court had a different attitude toward the case than had the appellant. In other words, the fact of prejudice is largely made to depend upon the errors which the appellant alleged are to be found in the record. To agree with the appellant it would be necessary to accept his theory of the case and find that the complainants were not justified in filing their suit; but our inspection of the whole record convinces us that they *239were, and under these circumstances we are of the opinion that the judgment appealed from should he

Affirmed.

A motion for reconsideration having been filed, on July '21, 1916, Me. Justice Wole delivered the following opinion of the court:

A motion for reconsideration has been filed and various reasons therefor have been alleged. It is evident that this court made a verbal mistake in one part of its opinion. We said in considering the tenth exception that the court below committed no error in excluding a letter from Víctor Martinez to his wife wherein he said that his only legitimate son was Víctor P. Martínez. The use of the word “legitimate” was a slip of the pen. It is evident that the purpose of the defendant, as revealed by the letter, was to show that he was the “only son,” not the “only legitimate son.” The letter was still no less in the nature of a self-serving declaration, no less res inter alios acta, and of practically no probative force. The only letter to which the tenth exception related was one directed to the wife of Víctor Martinez and not to a letter written to the Royal Bank of Canada. But a letter written to the hank would be subject to similar objections.

With respect to the proof that the complainants were the children of Cecilia Méndez and were horn on the dates mentioned, we have nothing to add over the reasoning of the opinion; nor have we anything to add with regard to the probative force of the letters admitted in evidence.

With respect to the interest of the witnesses or the contradiction in the proof, that was for the court below, and this court will not interfere except in an extraordinary case and when it is convinced of error or other element.

*240If the judgment of the court below depended solely on. the voluntary acknowledgment by Víctor Martinez of the alleged natural children, we should have considerable sympathy with the position of the defendant, but the judgment of this court is specifically based on the fact that the proof tended to show a state of concubinage existing between Victor Martinez and Cecilia Méndez and that the children were born during the continuance of that state. It was unnecessary, therefore, to prove anything more than that at the date of the birth of each of the children such a state of concubinage existed; and the proof, in fact, tended to show a state of concubinage beginning before and extending beyond the birth of either of the children in this case.

The motion must be

Overruled.